Safety and productivity are low at night and this would appear to be because we are a diurnal species. This is reflected not only in our habitual sleep time, but also in our endogenous body clocks that, together with exogenous influences, such as the patterning of meals and activity, result in predictable circadian (24 h) rhythms in our physiological processes. Our performance capabilities also vary over the course of our waking period, with task demands affecting both the precise trend over the day, and the rate at which it adjusts to the changes in sleep timing occasioned by shift work. Studies designed to examine the reasons for this have shown that memory loaded performance may have a quite separate endogenous component to that responsible for more simple performance, suggesting that these two types of performance cannot be causally related. Furthermore, it would appear that the exogenous component of circadian rhythms may also differ across measures, and our attempts to model these endogenous and exogenous components have led us to re-examine the evidence on adjustment to night work. Our findings suggest that shiftworkers merely 'stay up late' on the night shift, rather than adjust to it, and that this is responsible for the reduced safety at night. It would seem that in situations where safety is paramount, the only solution to these problems is the creation of a nocturnal sub-society that not only always works at night but also remains on a nocturnal routine on rest days.