The risk of developing a second primary cancer was evaluated in over 250,000 persons reported to the Connecticut Tumor Registry (CTR) during 1935-82. The CTR has collected data on cancer incidence longer than any other population-based tumor registry and thus provided researchers with a unique opportunity to investigate the occurrence of second cancers among persons followed for long periods, in some cases for more than 40 years. When compared with the general Connecticut population, cancer patients had a 31% increased risk of developing a subsequent cancer overall and a 23% elevated risk of second cancer at a different site from the first. Little variation in risk was seen for the first 20 years of follow-up, although the risk for females averaged twice that for males (41% vs. 18%). Persons who survived more than 20 years after the diagnosis of their first cancer were at highest risk: 51% for females and 45% for males. Over 1 million person-years of observation were recorded, and the excess risk of developing a new cancer was 3.5 per 1,000 persons per year. Common environmental exposures seemed responsible for the excess occurrence of many second cancers, particularly those related to cigarette smoking, alcohol consumption, or both. For example, persons with epithelial cancers of the lung, larynx, esophagus, buccal cavity, and pharynx were particularly prone to developing new cancers in the same or contiguous tissue throughout their lifetimes. A notable finding was the high risk of cancers of the lung, larynx, buccal cavity, and pharynx observed among cervical cancer patients, which suggested a common etiology involving cigarette smoking. The intriguing association previously reported among cancers of the colon, uterine corpus, breast, and ovary was confirmed in our data, which indicated the possible influence of hormonal or dietary factors. Incidental autopsy findings were largely responsible for the observed excesses of second cancers of the prostate and kidney, and heightened medical surveillance of cancer patients likely resulted in ascertainment bias and elevated risks for some tumors during the early period of follow-up, most notably cancers of the thyroid. Interestingly, patients with prostate cancer were the only ones found to be at significantly low risk for second cancer development. However, this might be an artifact of case-finding because advanced age at initial diagnosis of prostate cancer was associated with an underascertainment of second cancers.(ABSTRACT TRUNCATED AT 400 WORDS)