Final MIAB report on Bayesian yacht disaster reveals crew were unaware of critical safety risk

3 images

Credit: Perini Navi

Breaking: Final MAIB report on Bayesian yacht disaster reveals crew were unaware of critical safety risk

19 November 2026 • Written by Emily Dawkins

BOAT International has followed the Bayesian story since 19 August 2024, when the 56-metre Perini Navi was hit by bad weather and sank off the coast of Sicily. Discover our live blog coverage, where we continue to report on the aftermath of the tragedy as it unfolds. 

Read More/All the latest news surrounding the 56m sunken superyacht Bayesian as investigations continue

The Marine Accident Investigation Branch (MAIB) has published its final accident report on the sinking of Bayesian, more than two years after the disaster that tragically claimed the lives of seven people.

According to the 145-page report, the investigation ultimately found that, with no sails set, Bayesian’s windage and stability characteristics left her "vulnerable to knockdown in severe winds". It says that the yacht's loading condition, apparent wind angle, and centreboard position affected this vulnerability, though the findings now reveal that those possible weaknesses were not identified in the originally approved stability information provided to the crew. 

The report concluded that the information supplied in guidance to the mariners "did not provide sufficient detail of the impact of possible weather conditions to support decision-making".

MAIB also found that the weather forecasts for the area, while broadly accurate, "did not indicate the risk of downbursts from the thunderstorms predicted". 

Bayesian's salvage concluded on 30 June 2025
Credit: TMC Marine

What were the key analyses identified in the report? 

Regulation

Bayesian was constructed to meet LY2 requirements, which allow a sailing yacht to have an angle of vanishing stability below 90 degrees. The "agreed operational criteria" – which were contained within the approved Stability Information Booklet (SIB) on board – related to the automatic sail monitoring and control system and its ability to shed the wind heeling load with the yacht under sail.

The MAIB quoted the Wolfson Unit study, which concluded that Bayesian, in its "probable loss condition" (similar to the motoring condition with sails down and the centreboard in the raised position), was vulnerable to winds on the beam, with the steady-state wind heeling limit and the angle of vanishing stability in this condition lower than when in the sailing condition. 

Reference was made to the similarities of the Concordia (2010) and Pride of Baltimore (1986) accidents, in which wind heeling levers were a factor. Both LY2 and the later Red Ensign Group Yacht Code (REG Yacht Code) required stability curves and wind-heeling limits to be provided, but only in the sailing position. Neither code required assessing monohull sailing yachts like Bayesian without sails set, "placing the people on such vessels at an undetermined risk," the report stated. 

In addition, the report documents that Bayesian's SIB carried only generic storm advice and did not include information on the effect of the weather at anchor (in the motoring position with the centreboard raised and no sails set) - something that it noted "had never been assessed". It states that the crew were therefore "unaware of a critical risk to the vessel" in this position and were "unable to take account of it in their operations". 

Weather information

Official forecasts for the night from Météo-France and CNMCA predicted thunderstorms and winds up to force five, with gusts up to 40 per cent stronger across the Lipari sea area and Tyrrhenian Sea coastal regions. Though, according to the report, the forecasts did not detail the strength or severity of the gusts. 

Overall, the MAIB found that, although broadly accurate, the forecasts "did not communicate the potential severity and impact of the predicted thunderstorms", and that none explicitly warned the crew of Bayesian about the risk of extreme weather and the possibility of hurricane-force winds. The report also noted low-probability, high-impact events as "a significant challenge to all weather forecasters."

The site of the disaster, 0.5 nautical miles south-east of Porticello, Italy
Credit: Guardia Costiera Porticello

Guidance and training

MAIB also criticised UK Hydrographic Office publications, which offered "insufficient information for thunderstorms, waterspouts, downbursts, downdrafts or mesocyclonic supercells".  According to the report, the syllabus for masters of large sailing yachts under MSN 1858 (M) Amendment 1 also lacked sufficient detail on assessing risk near severe thunderstorms. The report says the Concordia investigation had flagged a similar knowledge gap. It noted that mountains, warm seas and forecast thunderstorms were all clues to local effects, but this wasn't covered in official guidance or training.

Life rafts

According to the report, none of Bayesian's life rafts floated free or inflated automatically when the yacht foundered, with a crew member having to release one manually at the time of the disaster. The rafts' hydrostatic release units all operated, but the stowage held the rafts on board. LY2 required float-free release but "did not adequately account for the knockdown and sinking scenario", the MAIB said, with the REG code requiring it only with the vessel upright. However, the report classifies this factor as not directly contributory.

Other points

LY2 rules set a 40-degree minimum downflooding angle, but the SIB listed openings immersing below that. The report noted that Bayesian "did not meet the LY2 requirements" at approval, though it classifies this as non-contributory. The MAIB also referenced the crew's reaction to the event, noting that "all those on board reacted to the situation they found themselves in, and any attempt to initiate the emergency procedures became superfluous".

According to the report, the crew needed to respond to a rapidly developing situation that had not been anticipated in existing onboard drills, noting that survivors' safety "relied heavily on the training, experience and initiative of individual crew members."

What are the recommendations made?

In light of the final report, MAIB has prepared a paper to submit to the International Maritime Organisation to highlight the investigation findings. Recommendations have been made to the following organisations: 

  • To the Red Ensign Group, to undertake a study of the vulnerabilities identified in the report and amend the Red Ensign Group Yacht Code and associated guidance.
  • To the Maritime and Coastguard Agency, to amend its training syllabus for yacht masters to include the risks from downbursts and severe wind gusts.
  • To the UK Met Office, to raise with the World Meteorological Organisation the need to consider how low-likelihood, high-impact weather hazards should be represented in its data services.
  • And finally, to the UK Hydrographic Office, to develop and update guidance on the risks of such extreme weather in its publications.

"Our thoughts remain with the families and loved ones of the seven who lost their lives, and with all those affected by the loss of Bayesian," said MAIB's chief inspector of marine accidents, Rob Loder. "Our investigation found that Bayesian had a vulnerability to high winds that was not recognised, because the yacht had never been assessed at anchor with its sails down. The extreme winds that night were also not explicitly forecast, so the crew had no warning of the risk."

He continued: "MAIB has published a safety flyer for the industry, and a report which makes recommendations covering vessel design standards, crew training and how rare but severe weather is represented in forecasts. I urge everyone with a part to play, including regulators, training providers and weather services, to read our report and act on it, so that a tragedy like this is not repeated."

The criminal investigation led by the Termini Imerese Public Prosecutor's Office in Sicily remains underway.

Read More/Why are there two investigations into Bayesian – and could they reach different conclusions?

Sponsored listings