Do ships and aircraft disappear unusually often in the Bermuda Triangle?

The Claim

Ships and aircraft disappear or experience unexplained losses in the Bermuda Triangle at a rate higher than comparable heavily travelled ocean regions.

The question is not whether vessels or aircraft have gone missing there. It is whether their risk of disappearance, relative to how much traffic passes through, is demonstrably higher than in suitably comparable regions. Maritime and aviation risks must be assessed separately.

Where It Comes From

The Triangle is commonly drawn between Florida, Bermuda and Puerto Rico, though its boundaries have never been standardized as an official navigational region. Popular books, magazines and retellings assembled incidents from different periods into one recognizable mystery.

Some of these losses were real, and some records remain incomplete. But an accumulation of prominent stories is not the same thing as an incidence-rate analysis.

Why It Sounds Convincing

A ship can be lost in deep water without recoverable wreckage. An aircraft can disappear over open ocean with little physical evidence. Dramatic cases such as Flight 19 become easier to remember than the much larger number of uneventful crossings.

When investigators add up losses but cannot count all comparable crossings, a busy and hazardous region can appear statistically exceptional even if its per-journey risk is not.

The Short Answer

A higher disappearance rate in the Bermuda Triangle has not been demonstrated. NOAA says there is no evidence that mysterious disappearances occur more often there than in other large, well-travelled ocean areas.

But an absence of demonstrated excess is not a measured finding of equal or lower risk. A frequently cited casualty-record study offers useful incident descriptions but lacks the shipping and aviation traffic denominators needed to settle that comparison. The alleged excess of disappearances is not supported by an adequately normalized comparison, and the true comparative rate remains unresolved.

How It Would Have to Work

To test a geographic anomaly, investigators would need a pre-defined Triangle boundary and matched comparison regions; consistent definitions of a disappearance, accident or unexplained loss; a complete enough record of events; and exposure measures such as ship-hours, voyages or nautical miles for maritime traffic and flight-hours or traversals for aviation.

The relevant comparison is the number of consistently defined losses divided by the traffic at risk in each region. Weather, vessel class, flight type, reporting jurisdiction, search-and-rescue records and changes over time would also need adjustment. A valid study would report uncertainty around the resulting rate ratio.

What Actually Happens

The surrounding Atlantic and Caribbean contain heavy traffic, hurricane-prone waters, the Gulf Stream, reefs, shallow banks and periods of dangerous weather. These conditions can help explain individual accidents, although any particular historical loss must be examined on its own evidence.

Conventional hazards do not prove that every disputed loss is solved. Nor would discovering one unexplained event demonstrate a distinctive regional hazard.

What the Evidence Shows

Maritime records. A 2022 working paper, described in a public methods account associated with its authors, considered 9,096 selected US Coast Guard-investigated offshore marine incidents from 1982 to 2015. Of those, 361 lay inside the authors’ chosen Triangle boundary, and none of the selected cases inside was coded as a “Disappearance.” This is not evidence that no ship ever disappeared there: records with missing or unsuitable location data and incidents within 80 miles of a coast were excluded, and the source concerns a specific investigation dataset.

Crucially, the authors state they could not obtain the maritime traffic-density data that would allow incidents to be normalized by vessel exposure. Their maps show locations and distributions of observed casualty records, not risk per voyage or per ship-hour. Comparing incident density on a map to a matched risk rate would be a measurement error.

Aviation records. In the authors’ separate U.S.-investigated aviation accident data, 53 recorded events occurred inside the chosen boundary. Forty-two concerned take-off or landing; among the 11 en-route incidents, two personal flights were recorded as missing. Without flight volumes or hours within each region, those numbers cannot establish either a higher or a lower disappearance rate per flight. The maritime and aviation observation periods also differ, preventing a pooled, common-period rate.

Institutional assessment. NOAA’s fact sheet says there is no evidence of an unusually high frequency of mysterious disappearances. This is relevant expert context, not a published matched-exposure dataset. It should not be represented as one.

The original SSRN PDF’s complete figures, underlying USCG/NTSB cases and analysis code were not independently reproduced in this review. The accessible author-associated methods account increases visibility into the numerators but does not remove the missing-denominator problem.

The Strongest Evidence for the Claim

Advocates point to famous unresolved losses, the difficulty of recovering wreckage from deep water and unusual testimony from some incidents. These are legitimate reasons to investigate an individual event carefully; they can also motivate systematic collection of better incident records.

To support this Claim, however, even a well-documented dramatic loss would need to be placed in a complete, consistently defined count and measured against the traffic exposed in comparable waters or airspace. An anecdote cannot supply that comparison.

The Strongest Evidence Against It

No validated exposure-adjusted regional excess was identified in the reviewed sources. NOAA’s institutional assessment says the available evidence does not indicate one, and the authors’ incident-record work does not reveal a striking concentration of coded disappearances in its selected offshore sample.

Those observations challenge the popular inference, but neither constitutes a controlled statistical test demonstrating identical or lower loss rates. That limitation prevents treating the original claim about elevated regional risk as decisively disproved.

Common Viral Arguments

“Dozens of ships disappeared there.” A raw total is not a probability. We need matching counts of the vessels or journeys that could have been lost.

“The study found no disappearances in the Triangle.” The paper reported no records classified as disappearances within its selected offshore USCG sample. That narrow result does not describe every ship loss or quantify regional risk.

“Scientists proved the Triangle is completely normal.” The evidence does not establish an anomalous higher rate, but a sufficiently controlled marine-and-air comparison has not been presented here. “Not demonstrated” is not equivalent to “proved impossible.”

“Compasses behave mysteriously there.” Navigation difficulties and magnetic variation are not, without further data, evidence for a higher disappearance rate per crossing.

The Reasoning Error

Base-rate neglect mistakes a selected numerator—memorable losses—for the rate of loss among all exposed voyages and flights. Anomaly hunting then concentrates attention on events assigned to the Triangle while ignoring ordinary crossings, omitted cases and comparison regions. Cherry-picking can make an inconsistent historical collection appear like a single coherent statistical sample.

The correction applies the same discipline in both directions: lack of proof of an excess is not proof of no excess.

What Would Change Our Conclusion?

A transparently assembled, geocoded and independently reproducible series of ship and aircraft disappearances, combined with time-matched traffic exposure and well-matched comparator regions, could establish or rule out a meaningful excess within reported uncertainty. Marine and aviation rate ratios would need separate definitions and estimates.

An independently verifiable new study measuring a genuine excess would strengthen the claim. A robust matched-exposure analysis ruling one out could support a stronger negative assessment than the current evidence permits.

Evidence Status

Status: Unsupported
Evidence quality: Low

The proposed regional higher risk has not been demonstrated with reliable denominators and valid comparison cohorts. Low describes the directness and precision of the available evidence about comparative geographic risk; it does not mean that the existence of storms, currents or ordinary navigation hazards is doubtful.

If the question were to estimate the actual size or direction of any difference in regional risk, the answer would remain inconclusive without reliable traffic denominators. The published assessment instead addresses the original positive claim of elevated risk, for which adequate supporting evidence has not been demonstrated; the true comparative rate remains unresolved.

Evidence & Sources

The linked Evidence records separate (1) counts and classifications of selected marine and aviation incidents, (2) the original authors’ admitted inability to obtain traffic exposure, and (3) NOAA’s broader assessment. None of these sources is presented as a replicated, exposure-normalized regional disappearance-rate experiment.

Corrections & Updates

October 15, 2026 — Corrected the Bermuda Triangle evidence status from Contradicted / Moderate to Unsupported / Low: the 2022 casualty study lacked shipping and flight-exposure denominators and cannot establish a matched regional disappearance-rate comparison.

Last Reviewed

October 15, 2026.