Over the last decade, the shipping industry has implemented a number of measures aimed at improving its safety level (such as new regulations or new forms of team training). Despite this evolution, shipping accidents, and particularly collisions, remain a major concern. This paper presents a modified version of the Human Factors Analysis and Classification System, which has been adapted to the maritime context and used to analyse human and organisational factors in collisions reported by the Marine Accident and Investigation Branch (UK) and the Transportation Safety Board (Canada). The analysis shows that most collisions are due to decision errors. At the precondition level, it highlights the importance of the following factors: poor visibility and misuse of instruments (environmental factors), loss of situation awareness or deficit of attention (conditions of operators), deficits in inter-ship communications or Bridge Resource Management (personnel factors). At the leadership level, the analysis reveals the frequent planning of inappropriate operations and non-compliance with the Safety Management System (SMS). The Multiple Accident Analysis provides an important finding concerning three classes of accidents. Inter-ship communications problems and Bridge Resource Management deficiencies are closely linked to collisions occurring in restricted waters and involving pilot-carrying vessels. Another class of collisions is associated with situations of poor visibility, in open sea, and shows deficiencies at every level of the socio-technical system (technical environment, condition of operators, leadership level, and organisational level). The third class is characterised by non-compliance with the SMS. This study shows the importance of Bridge Resource Management for situations of navigation with a pilot on board in restricted waters. It also points out the necessity to investigate, for situations of navigation in open sea, the masters' decisions in critical conditions as well as the causes of non-compliance with SMS.
