Physical and social characteristics recorded at college physical examination and reported in subsequent questionnaires to alumni in 1962 or 1966 by 50,000 former students from Harvard University and the University of Pennsylvania were reviewed for their relationship to major site-specific cancer occurrence. The records of 1,359 subjects who died with a major site-specific cancer in a 16- to 50-year follow-up period and of 672 subjects who reported such a cancer by mail questionnaire in 1976 or 1977 were compared with those of 8,084 matched classmates who were known to be alive and free of cancer at the time subjects with cancer had died or had been diagnosed. Cigarette smoking, as reported both in student years and years as alumni, predicted increased risk for cancers of the respiratory tract, pancreas, and bladder. Student coffee consumption was associated with elevated risk for leukemia, but it was unrelated to cancers of the pancreas and bladder. Male students with a record of proteinuria at college physical examination experienced increased risk of kidney cancer, and those with a history of tonsillectomy experienced increased risk of prostate cancer. Students who at college entrance reported occasional vague abdominal pain were at elevated risk for pancreatic and colorectal cancers in later years. Increased body weight during college was associated with increased risks of kidney and bladder cancers, whereas for alumni this index was associated only with kidney cancer. Increased weight-for-height during college (but not in 1962 or 1966) predicted increased occurrence of female breast cancer. Jewish students experienced elevated risk for subsequent cancers of the female breast, colon, and combined colorectum. These and other findings are presented as clues deserving further exploration for any etiologic significance that they may hold for the cancer sites studied.