In his review Wilkinson considers the question 'How fast should the night shift rotate? He concludes quite strongly that with the possible exception of social factors, most of the evidence favours the use of prolonged or permanent night shifts. In this reply I argue that in reaching this conclusion Wilkinson has (i) overestimated the problems associated with rapidly-rotating shift systems; and (ii) underestimated the problems that might be encountered in trying to implement effective permanent night shift systems. The evidence on the various problems reviewed by Wilkinson is reconsidered. Other aspects of shift systems that were ignored by Wilkinson are then also taken into account. It is concluded that the evidence does not allow a general choice to be made between the use of either permanent night shifts or rapidly rotating, delaying, shift system. Rather the choice between them would appear to depend on the relative importance attached to safety and social problems in any given workplace.