This study aimed to determine whether there was a peak in accidents in the second to fourth hours within a shift and to compare the effect on accidents of the presence or absence of a break of only eight hours between shifts in a shift cycle ("quick returns"). Archival accident records were obtained for a four-year period from two sister works within the same division of a heavy-industrial organization. The works had many similarities but differed in the shift systems used. One had "quick returns," whereas the other did not. Accident frequencies were analyzed across four variables, site, shift, time on shift, and process code (process code identified the process area in which the worker operated). These were analyzed by hierarchical log-linear analysis, using a significance level of 0.05. The results indicated there was a peak in accidents in the second and third hours into a shift, but provided only slight evidence for a detrimental effect of "quick returns."