Abstract
Reviewing the literature on time on task effects on safety shows contradictory evidence, especially with regard to 12h shifts. It is argued that this might depend on methodological problems associated with the analysis of accident data, e.g. selectivity of samples, validity of data bases and study designs, especially for analyses at the company level. Analyses of aggregated data seem to indicate an exponential increase of accident risk with time on task beyond the normal working day. This is supported by some recent studies based on data from the Federal Republic of Germany.
