The incidence of cancer at sites other than the testis has been investigated in the families of 797 Norwegian and 178 Swedish patients diagnosed with testicular cancer during 1981-91. In the families of the Norwegian patients, the total number of cancers in the relatives was significantly lower than the expected number derived from national incidence rates [observed number of cancers 250, expected number of cancers 281.92, standardised incidence ratio (SIR) 0.89, 95% confidence interval (CI) 0.78-1.00]. This finding can be accounted for almost entirely by the finding of fewer than expected prostate and gastrointestinal cancers in the parents of cases. The other common cancers were found at slightly lower than or near the expected levels in the relatives. In the Swedish cohort, which accounts for less than 20% of cases, the observed number of cancers was very close to the expected number. Fourteen fathers of cases had prostate cancer compared with 27.57 prostate cancers expected, giving a SIR of 0.51 (P=0.006). Mothers had more lung cancers (ten cases observed, SIR=2.11, P=0.04) and cancers of the endometrium than expected (13 cases observed, SIR=1.73, P=0.09). These findings may be interpreted as support for theories proposing hormonal dysfunction as causing testicular cancer. Fifty-four gastrointestinal cancers were observed in the parents compared with 68.48 expected (SIR=0.78, P=0.082). Furthermore, testicular cancer was not found to be associated with the known dominantly inherited cancer syndromes [Familial breast (-ovarian) cancer, hereditary no-polyposis colon cancer]. However, one patient belonged to a Li-Fraumeni family, raising the possibility that testicular cancer may be an infrequent component of this rare cancer syndrome. This study supports the hypothesis that families of testicular cancer patients are not prone to cancer.