Document Type

Article Restricted

Publication Date


Journal Title

Accident Analysis & Prevention

Volume Number


Issue Number


First Page


Last Page



The majority of current maritime regulations has been developed following a reactive approach, often as ad-hoc response to serious accidents, and are characterised as being prescriptive leaving limited space for adapting equivalent solutions rather those described in the regulations. On the premise of providing a more proactive approach for the proposal or the evaluation of regulations, the Formal Safety Assessment (FSA) has been introduced. In the context of FSA, the analysis of accident data is considered to be very important for providing potential input on developing more balanced, proactive and cost-effective regulations. However, it has been argued that the validity of historical data may be undermined by uncertainties. This paper is aimed at showing evidence on serious under-reporting in accident databases, which can be considered as the main contributor to questioning the direct and uncritical use of historical data. By analysing the 10-year tanker accident data from the Lloyd's Register FairPlay (LRFP) and the Norwegian Maritime Directorate (NMD) for vessels registered in Norway, it is found that the reporting performance has an upper bound of 41% for NMD and 30% for LRFP. Furthermore, based on comparison between LRFP data and self-assessment by Flag States, it is seen that accidents reported by the Flag States are also incomplete.