USCG Marine Casualty Data: Every US Vessel Accident Since 1982

The US Coast Guard operates two parallel marine accident databases that together document every significant vessel casualty in American waters: the Boating Accident Report Database, covering recreational boating accidents reported by states since the 1960s, and the Marine Casualty and Pollution Database, covering commercial vessel casualties since 1982. Between them the databases record tens of thousands of incidents annually—capsizings, groundings, collisions, fires, sinkings, and alcohol-related fatalities on recreational craft; machinery failures, personnel casualties, and hazardous material pollution events on commercial vessels. The data is underused by researchers and routinely mined by litigators, investigative journalists, and safety advocates. Here is what each database contains, how to access it, and what the records reveal.

The two databases: BARD and MCPD

The Coast Guard's accident data infrastructure reflects the fundamental regulatory divide between recreational and commercial maritime activity. Recreational boating regulation is largely delegated to the states; the Coast Guard sets federal standards for required safety equipment, numbering, and operator conduct, but accident reporting flows from states to the federal level rather than directly from operators. Commercial vessel regulation is primarily federal, with the Coast Guard exercising direct inspection and reporting authority over vessels engaged in commercial operations.

BARD — Boating Accident Report Database is the national repository for recreational boating accident reports submitted by state boating law administrators to the Coast Guard under 46 U.S.C. § 6102. Every state and US territory with a recreational boating program is required to collect accident reports from operators and forward aggregate data to the Coast Guard annually. The national dataset begins in the late 1960s, though the reporting format has been standardized and the data is most analytically consistent from the early 1980s onward. Each BARD record documents a single recreational boating accident event with fields covering vessel characteristics, operator demographics and sobriety, environmental conditions, incident type, primary cause, and casualty counts.

MCPD — Marine Casualty and Pollution Database covers commercial vessel casualties reported directly to the Coast Guard under 46 CFR Part 4 and marine pollution incidents reported under 33 CFR Part 153. Commercial vessel operators are required to report casualties that result in: the death of any person; the injury of any person that requires professional medical treatment beyond first aid; the loss of a vessel; damage to a vessel or property exceeding a dollar threshold (currently $25,000); or a significant pollution event affecting the marine environment. MCPD records are more detailed than BARD records—they include vessel documentation numbers, Coast Guard Marine Inspection Office identifiers, full investigation findings, and pollution quantity estimates—and are maintained within the Marine Information for Safety and Law Enforcement system, known as MISLE.

BARD data structure: what each recreational accident record contains

BARD records are submitted by state boating law administrators following their collection from accident operators (who are legally required to file accident reports with their state agency when a qualifying accident occurs). The key fields:

• Vessel type and propulsion — classification of the vessel involved by hull type (open motorboat, cabin motorboat, personal watercraft, rowboat, canoe, kayak, sailboat, inflatable, and others) and propulsion type (outboard, inboard, inboard/outboard, jet drive, paddle, sail, or none). Powered recreational vessels—motorboats and personal watercraft—account for more than 75 percent of BARD incidents and an even higher share of property damage, though paddlecraft contribute disproportionately to fatality counts relative to incident counts because capsize events on rivers and lakes in non-powered craft frequently result in drowning.
• Operator age and experience — age of the vessel operator at the time of the accident, and total years of boating experience. The operator age distribution in BARD fatality records is not concentrated among the youngest operators (as it is in motor vehicle crash data); older operators in the 40–60 age range account for a substantial share of fatal accidents, partly because they represent a large share of vessel ownership. Experience, measured in years, is a weaker predictor of accident risk than license status (in states with mandatory boater education requirements) and alcohol involvement.
• Operator alcohol involvement — a flag and categorical field indicating whether alcohol was involved in the accident and, where tested, the blood alcohol concentration of the operator. Alcohol is the operator-reported primary contributing cause in more than 15 percent of annual boating fatalities in the BARD dataset—a figure that understates the true proportion because alcohol involvement in fatal accidents involving drowning is frequently not tested or reported. The Coast Guard and National Association of State Boating Law Administrators consider alcohol the single most consequential controllable risk factor in recreational boating fatalities.
• Primary cause — the operator or state agency's determination of the primary contributing cause of the accident. The BARD cause taxonomy includes: operator inattention (the most common single cause code across all incident types), operator inexperience, improper lookout, excessive speed, alcohol or drug use, machinery failure, weather, navigation rules violation, and striking a submerged object. Collision with another vessel is consistently one of the top primary cause categories by incident count.
• Incident type — the physical event that constituted the accident: collision with another vessel, collision with a fixed object, capsizing, flooding or swamping, grounding, fire or explosion, fall overboard, skier accident, and others. Capsizing and flooding or swamping are among the leading incident types in BARD fatality records because they result in immersion, which produces drowning risk that is dramatically higher in the absence of a life jacket.
• Life jacket use — whether the victim was wearing a personal flotation device at the time of the accident. This field consistently documents one of the most striking patterns in BARD: in the large majority of drowning deaths occurring in recreational boating accidents, the victim was not wearing a life jacket. The Coast Guard's annual Recreational Boating Statistics report documents this consistently—in most recent years, more than 80 percent of drowning victims in recreational boating fatalities were not wearing a life jacket. Life jacket non-use is the dominant preventable factor in boating drowning deaths, more statistically prominent than alcohol involvement.
• Fatalities, injuries, and property damage — the reporting threshold for BARD requires state submission for any accident involving death, injury requiring medical treatment, disappearance of any person from a vessel under circumstances suggesting death or injury, or damage to vessels or property exceeding $2,000. The $2,000 property damage threshold is low relative to actual boat repair costs, meaning a substantial share of BARD records involve property-damage-only accidents. Fatality records are the most analytically complete portion of the database; minor property damage incidents may be significantly underreported in states with lower enforcement of accident reporting requirements.

MCPD data structure: commercial vessel casualties

The Marine Casualty and Pollution Database covers a fundamentally different population than BARD: vessels engaged in commercial activity, from large oceangoing cargo ships and tankers to small passenger ferries, fishing vessels, and towing vessels. The reporting obligation falls on the vessel owner or operator, not a state agency, and the Coast Guard conducts its own investigations of significant casualties rather than relying solely on self-reporting. Key MCPD fields include:

• Vessel documentation and type — commercial vessels operating on navigable waters are documented under 46 U.S.C. Chapter 121 and carry a unique official number issued by the Coast Guard. The documentation number is the primary vessel identifier in MCPD and links to the Coast Guard vessel documentation database, which records ownership, tonnage, vessel type, and hailing port. Vessel types in MCPD span the full range of commercial maritime: cargo vessels, tank vessels (oil and chemical tankers), towing vessels (tugboats), fishing vessels, passenger vessels (ferries, excursion vessels, dinner boats), offshore supply vessels, and mobile offshore drilling units.
• Casualty type — the physical event category: grounding, allision (striking a fixed object), collision with another vessel, sinking or foundering, flooding, fire or explosion, machinery failure or loss of propulsion, structural failure, personnel casualty (injury or death without vessel damage), and pollution without vessel casualty. Groundings and machinery failures collectively account for a large share of commercial casualty events by count; fires and sinkings account for a disproportionately large share of fatalities and major property losses.
• Personnel casualties — deaths and injuries to crew, passengers, and others attributable to the casualty event. Commercial vessel personnel casualties are classified by whether the victim was a crew member, passenger, or person not on the vessel (a person in the water struck by a vessel, for example). Crew fatality rates in commercial fishing—documented in MCPD fishing vessel records—are among the highest occupational fatality rates of any industry in the United States.
• Pollution incidents — MCPD documents oil and hazardous material releases into navigable waters resulting from vessel casualties, including the type of pollutant, estimated quantity spilled, and whether a National Response Center report was filed under 33 U.S.C. § 1321. Pollution records in MCPD range from minor fuel oil spills during vessel groundings to major tanker casualties producing large environmental damage events. Incidents triggering Coast Guard cleanup response are linked to response cost records within MISLE.

Key findings from BARD: patterns in recreational boating fatalities

Decades of BARD data support several consistent findings about recreational boating risk that the Coast Guard highlights annually and that independent researchers have confirmed with more granular analysis.

Capsizing and flooding are the leading fatal incident types. While collision with another vessel is the most common incident type by total count, capsizing and flooding or swamping are disproportionately represented in fatal outcomes. The mechanism is straightforward: capsizing and flooding cause immersion, and immersion causes drowning in the absence of effective flotation. The combination of capsizing as the incident type with life jacket non-use as the victim status produces the majority of recreational boating drowning deaths in BARD.

Collision with another vessel dominates by incident count. Across the full BARD dataset, collision with another vessel is consistently the primary cause type by number of incidents, reflecting the concentration of recreational boating in congested waterways, the absence of formal traffic control on most recreational waterways, and the variable navigation rules training of recreational operators. Collisions with fixed objects—docks, pilings, rocks, and bridge structures—are a secondary collision category. Property damage is high in vessel-to-vessel collisions because both vessels sustain damage; fatalities are lower than in capsizing events because immersion is less likely.

Alcohol is the leading operator-attributed cause of fatalities. In years where alcohol involvement is explicitly coded in BARD, it is attributed as the primary cause in more than 15 percent of boating fatalities—a figure that the Coast Guard acknowledges is conservative. Unlike motor vehicle crashes, where blood alcohol concentration is routinely measured following fatalities, boating fatality investigations often lack toxicological testing because bodies are recovered from water and testing is delayed or not performed. The operational alcohol involvement rate in fatal boating incidents is estimated by researchers to be substantially higher than the BARD-documented 15 percent.

Powered recreational boats dominate the incident count. Open motorboats, cabin motorboats, and personal watercraft collectively account for more than 75 percent of all BARD incidents. This reflects both the large share of the registered vessel fleet that is powered and the higher speed and energy involved in motorized incidents. Personal watercraft—jet skis and similar craft—are overrepresented in injury incidents relative to their share of total boating activity, owing to their higher speed capability, frequently inexperienced operators, and the physical nature of falls and collisions at speed.

Seasonal and geographic concentration

BARD incidents are not uniformly distributed across the calendar year or US geography. The seasonal concentration of recreational boating activity drives a corresponding concentration of accidents. Memorial Day weekend through Labor Day weekend—the unofficial recreational boating season in most of the continental United States— accounts for the large majority of annual incidents. July is consistently the peak month for both total incidents and fatalities in BARD. The combination of maximum water traffic, warm-weather alcohol consumption, and extended daylight hours produces the seasonal peak.

Geographic concentration in BARD reflects the distribution of recreational boating activity and population. Florida consistently reports the highest absolute number of recreational boating accidents of any state, a consequence of its year-round boating season, large coastal and inland waterway network, large vessel registration count, and significant tourist boating activity. The Gulf Coast states—Texas, Louisiana, Alabama, and Mississippi—also report high incident counts. The Great Lakes region—Michigan, Wisconsin, Minnesota, Ohio, and New York—concentrates a large share of warm-weather boating activity and corresponding incidents into a shorter May–September season. Alaska, despite its relatively small population, contributes disproportionately to commercial vessel fatalities in MCPD because of the scale of its commercial fishing industry and the severe operating conditions in Alaska waters.

State-level BARD data requires normalization before comparison. A state with more registered vessels, more miles of navigable waterway, or a longer boating season will produce more absolute incidents without necessarily being more dangerous per boating hour. The appropriate denominator for interstate comparison is registered vessel count (published alongside BARD data in the annual Recreational Boating Statistics report) or, where available, estimated annual boating days derived from survey data.

How to access the data

The two databases have substantially different access characteristics that reflect their different administrative locations and data sensitivity.

BARD statistics at uscgboating.org. The Coast Guard publishes an annual Recreational Boating Statistics report aggregating BARD data at uscgboating.org/statistics/accident-statistics.php. The annual report is available as a PDF and in recent years as a downloadable Excel or CSV file. It provides tabulations by incident type, primary cause, vessel type, operator age, alcohol involvement, life jacket use, and state. The published statistics are aggregates, not individual incident records; individual-record BARD data is not publicly released because it contains personally identifiable information about operators and victims. The annual statistics files going back to 2000 provide sufficient data for longitudinal trend analysis at the national and state level.

MCPD via FOIA request or NTSB marine reports. The Marine Casualty and Pollution Database is maintained within MISLE, which is an internal Coast Guard law enforcement and marine safety system. MCPD records are not publicly searchable online; accessing specific commercial casualty records requires a Freedom of Information Act request to the Coast Guard Freedom of Information Act Division. FOIA requests for MCPD records covering specific vessels, time periods, or geographic areas are routinely processed and typically produce the Coast Guard investigation report, supplementary forms, and where applicable, pollution incident documentation. For the most significant commercial casualties, the National Transportation Safety Board's Marine Accident Reports—published as full public dockets at ntsb.gov—contain the MCPD record as well as the NTSB's independent investigation findings and probable cause determination.

MISLE public query. The Coast Guard provides a limited public interface to MISLE inspection records (not casualty records) through the Port State Information eXchange tool (PSIX) and the Vessel Inspection Queries portal, which allows lookup of vessel inspection history by vessel name or IMO number. These tools cover Coast Guard safety examinations and deficiency findings, not casualty records, but they are useful for researching a specific commercial vessel's inspection history prior to a casualty event.

The NTSB marine investigation relationship

The National Transportation Safety Board conducts independent investigations of major marine accidents under 49 U.S.C. § 1131. The NTSB's marine accident investigation authority is concurrent with the Coast Guard's—both agencies may investigate the same accident—but the NTSB's role is to determine probable cause and issue safety recommendations rather than to assess regulatory violations or initiate enforcement actions. For the most significant commercial vessel casualties—large passenger vessel accidents, casualties involving multiple fatalities, accidents with significant policy implications for maritime safety regulation—NTSB investigations produce the most analytically complete public record.

NTSB marine accident investigation dockets are published in full at ntsb.gov. Each docket for a major marine accident includes: the full investigation report with factual findings and probable cause determination; factual reports from NTSB investigators (covering vessel characteristics, environmental conditions, voyage data recorder output where available, crew interviews, and post-accident inspection findings); exhibits including voyage track charts, vessel photographs, and laboratory analysis of failed components; and safety recommendations issued to the Coast Guard, flag state, vessel operators, or industry associations. NTSB marine investigation dockets are the closest equivalent for marine accidents to NTSB aviation investigation dockets—they are comprehensive, publicly accessible, and analytically richer than the underlying MCPD record alone.

The NTSB issues safety recommendations to the Coast Guard following major marine investigations, and the Coast Guard's response status to open recommendations is tracked on the NTSB safety recommendation database. Tracking open marine safety recommendations—those classified as “Open—Unacceptable Response” or “Open—Await Response”—provides a view into which structural maritime safety problems have been identified by investigators but not yet addressed by the regulatory agency.

Analyzing BARD annual statistics with Python

The Coast Guard's publicly available annual statistics files are the most practical starting point for BARD analysis. The following script demonstrates loading and analyzing the aggregate tabulations published in recent annual reports:

import pandas as pd

# USCG publishes annual Recreational Boating Statistics reports as PDFs.
# The underlying state-reported BARD statistics are also released as
# structured tables in those reports. This example uses the summary CSV
# files extracted from recent annual reports, available at:
#   https://www.uscgboating.org/statistics/accident-statistics.php
#
# Download the Excel/CSV version of the annual statistics for each year
# and concatenate across years for longitudinal analysis.

years = list(range(2000, 2024))
frames = []

for year in years:
    # Each file has consistent column naming from 2003 onward
    path = "bard_" + str(year) + ".csv"
    df = pd.read_csv(path, dtype={
        "state": str,
        "incident_type": str,
        "primary_cause": str,
        "vessel_type": str,
        "operator_alcohol": str,
        "fatalities": "Int64",
        "injuries": "Int64",
        "property_damage_usd": "Float64",
        "total_incidents": "Int64",
    })
    df["year"] = year
    frames.append(df)

bard = pd.concat(frames, ignore_index=True)

# Alcohol-related fatalities as share of total
alcohol_fatal = bard[bard["operator_alcohol"] == "Y"]["fatalities"].sum()
total_fatal = bard["fatalities"].sum()
print("Alcohol share of fatalities: " + str(round(alcohol_fatal / total_fatal * 100, 1)) + "%")

# Incident type breakdown
type_counts = (
    bard.groupby("incident_type")["total_incidents"]
    .sum()
    .sort_values(ascending=False)
)
print(type_counts.head(10).to_string())

# State-level fatality rate (fatalities per 100k registered vessels)
# Join to USCG registered vessel counts published in the same annual report
state_summary = (
    bard.groupby("state")[["fatalities", "total_incidents"]]
    .sum()
    .reset_index()
)

# Top 10 states by fatality count over the full period
print(state_summary.sort_values("fatalities", ascending=False).head(10).to_string())

# Seasonal concentration: incidents by month (where month data is available)
if "month" in bard.columns:
    monthly = bard.groupby("month")["total_incidents"].sum()
    peak_month = monthly.idxmax()
    print("Peak incident month: " + str(peak_month))

The script above illustrates the core analytical patterns for BARD aggregate data: computing alcohol's share of fatalities, ranking incident types by frequency, comparing state-level fatality counts, and identifying seasonal concentration by month. For more granular analysis requiring individual incident records rather than aggregate statistics, a FOIA request to the Coast Guard for BARD microdata (with PII redacted) is the appropriate path.

How journalists and litigators use the data

The Coast Guard's marine accident databases have proven useful across several categories of investigative and legal research, each exploiting a different dimension of the data.

Boat manufacturer defect tracking. BARD incident type and cause fields, combined with vessel make and model data in individual BARD records (accessible via FOIA), can surface patterns in accidents associated with specific vessel designs—hull instability in particular conditions, fuel system fires in specific engine configurations, or hatch failure patterns in flooding incidents. These patterns are difficult to establish from individual cases; the BARD population of accidents associated with a given make and model across multiple states and years is the necessary data for demonstrating a pattern. Product liability litigation involving recreational vessels routinely involves BARD data analysis as a component of expert testimony on whether a particular failure mode was novel or known and recurring in the population.

Rental company safety records. Commercial vessel rental operations—boat rental fleets, personal watercraft rental concessions, and guided excursion vessels—are subject to Coast Guard commercial vessel regulations when they carry passengers for hire. MCPD records for specific commercial operators can be obtained via FOIA and compared across operators of similar vessel types. Investigative journalists have used MCPD data to document patterns in excursion vessel operators with repeated casualty records—groundings, passenger injuries, or flooding events—that persisted across inspection cycles without generating enforcement actions proportional to the risk pattern.

Alcohol enforcement and deterrence research. The gap between alcohol's known role in boating fatalities and the low rate of boating under the influence arrests and prosecutions is a recurring policy research question. BARD data quantifies the outcome side of the equation—alcohol-attributed fatalities and incidents by state and year—while state arrest data from the Bureau of Justice Statistics provides the enforcement side. States with active BUI enforcement programs can be compared against states with minimal enforcement using BARD fatality rates as the outcome variable, controlling for registered vessel counts and boating season length. This type of natural experiment analysis has informed Coast Guard and NASBLA recommendations on BUI enforcement prioritization.

Commercial fishing vessel safety. MCPD records for fishing vessels document one of the most hazardous occupational sectors in the United States. The Fishing Vessel Safety Act of 1988 and its subsequent amendments expanded Coast Guard safety inspection requirements for commercial fishing vessels, and MCPD data from the pre- and post-Act periods provides the baseline for evaluating the regulatory intervention's effect on commercial fishing vessel casualty rates. Researchers at the National Institute for Occupational Safety and Health have used MCPD fishing vessel records extensively in occupational fatality rate studies for the commercial fishing industry.

Limitations of the USCG accident data

Both BARD and MCPD have structural limitations that shape what the data can support analytically. Understanding these limitations is prerequisite to drawing valid conclusions from either database.

BARD underreports minor incidents and varies by state. The $2,000 property damage threshold triggers a reporting obligation, but enforcement of the recreational boating accident reporting requirement varies substantially by state. States with active enforcement of the operator filing requirement, established relationships between marine patrol units and the boating accident report system, and electronic submission infrastructure report higher incident counts than states where reporting is largely voluntary in practice. This means that interstate comparisons of BARD incident counts conflate true differences in safety outcomes with differences in reporting completeness. Fatality records are the most reliable portion of BARD; deaths in boating accidents are difficult to conceal and are subject to medical examiner and law enforcement processes that generate records independently of operator reporting. Minor property damage incidents are the most underreported category.

BARD alcohol data is constrained by testing rates. The alcohol involvement field in BARD is populated from operator reports and, where performed, chemical testing. In fatal accidents involving drowning, the victim is recovered from water—often hours or days after the accident—and blood alcohol concentration testing is frequently not performed, not feasible, or not admissible for the purposes for which BARD records are used. The systematic undertesting of fatal drowning victims for alcohol means that BARD alcohol involvement statistics are a floor, not a ceiling, on the true alcohol-attributable fraction of boating fatalities.

MCPD access requires FOIA and processing time. Unlike BARD aggregate statistics, which are published annually as a matter of course, MCPD individual records are not publicly available through a self-service portal. Accessing specific commercial casualty records requires FOIA requests, which are processed with variable timeliness and frequently involve redactions for ongoing law enforcement or litigation sensitivity. For recently occurring casualties, the MCPD investigation record may not be finalized—the Coast Guard investigation process can take months for complex casualties—meaning the publicly accessible record will be incomplete until the investigation is closed.