The effects on industrial shift workers of the timing of the changeover from the night to morning shift and of the length of the shift were examined. Two groups changed over at 6 a.m. and worked either 8-hr or 12-hr shift systems; the other 2 groups changed over at 7 a.m., working either 8- or 12-hr systems. Night sleeps between consecutive shifts that started at 6 a.m. were shorter and more disrupted than those starting at 7 a.m. Day sleeps following night shifts that finished at 6 a.m. were longer and less disrupted than those finishing at 7 a.m. Early starts were associated with poorer psychological and physical health. These effects of changeover time are considered in relation to the circadian rhythms in sleep duration and propensity. Although several measures favored 12-hr shifts, physical health indicators appeared to favor 8-hr systems, especially in combination with late changeovers. However, the observed effects of shift length on chronic outcome measures are somewhat inconsistent with previous research findings.