Despite the increasing use of continuous process shift work in modern industry, few studies on the medical aspects of shift work can be found in recent literature of occupational health. Physiologists have shown that the ability of the body to adjust its circadian rhythms to alteration in hours of work or sleep can take up to a month. The usual type of shift work in industry involves weekly changes of hours, and thus on theoretical grounds at least this may not be the most suitable frequency for shift changes. Sickness absence of male refinery workers has been studied over a four-year period. The figures show that continuous three-cycle shift workers have consistently and significantly lower rates of sickness than day workers in similar occupations. The annual inception rate (spells) standardized for age was 108% for shift workers and 182% for day workers, and the average annual duration per man was 11 days for shift workers and 18 days for day workers, although the average length of spell was slightly longer among shift workers. As far as is known, such a difference has not been described before in detail. Age-related lateness and absenteeism have been measured and show similar wide differences between the two groups. Although both types of worker are largely self-selected, the difference is not due to medical selection or to an excess of any one type of disease in day workers. Over three-quarters of 150 shift workers interviewed stated that they preferred shift work hours and that sleeping difficulties were not common. It is suggested that the main reasons for the difference between shift and day workers9 sickness absence lie in the degree of personal involvement in the work and in the social structure of the working group.