A multiple-site case-control study of 15 cancers (stomach; colon; rectum; larynx; lung; melanoma; skin; female breast; male breast; cervix; ovary; uterus; prostate; testis; and bladder) was conducted to evaluate their association with occupational physical activity and socioeconomic status (SES). A hospital-based study population (3,486 male cases and 379 female cases, and 2,127 male and 244 female controls) was established in an oncological treatment center in Istanbul, Turkey, from 1979-84. Assessment of physical activity and SES was based on job titles held by the study subjects. Two measures of physical activity were developed based on energy expenditure and 'sitting time' during working hours. Observed risks were adjusted for age, smoking, and SES. Elevated risks were observed among workers who held sedentary jobs for cancers of the colon (odds ratio [OR] = 1.6), rectum (OR = 1.3), melanoma (OR = 1.9), male breast (OR = 1.4), prostate (OR = 5.0), and ovary (OR = 2.0). Cancers of the cervix and uterus showed significantly decreasing risks with decreased activity. Risks of cancers of the colon, rectum, larynx, ovary, and melanoma were enhanced after risks for physical activity indices were adjusted for SES, while the associations between physical activity and cancers of the prostate, cervix, and uterus were weakened after SES adjustment. Risks of melanoma rose significantly with both activity indices after SES adjustment.(ABSTRACT TRUNCATED AT 250 WORDS)