BACKGROUND: Because married couples share at least their home environment, spousal aggregation of cancer might provide clues to unsuspected etiologic factors. The authors sought to measure the concordance of cancer occurrence in married couples and explore factors that might explain greater-than-expected concordance. METHODS: The authors identified 25,670 cancer-free married couples in northern California who were followed for up to 31 years for the development of cancer. In Cox proportional hazards analysis, the development of cancer in a spouse was treated as a time-dependent, independent variable, and spouse-with/spouse-without risk ratios were determined, controlling for age and gender. For selected concordant espoused pairs, additional explanatory information was sought in their medical records. RESULTS: There was no excess concordance for all cancers combined; the spouse-with/spouse-without risk ratio was 0.97 (95% confidence interval, 0.90-1.05). Statistically significant husband-wife associations were found only for cancer of the tongue and stomach and for non-Hodgkin lymphoma. Except for cancer of the penis/endometrium and testis/vulva, based on one couple with each combination, gender specific cancers did not aggregate within married couples. Established and suspected risk factors, not necessarily related to the marriage, were found for some individuals who had concordance with their spouses. CONCLUSIONS: Little spousal concordance for cancer occurrence was found. The study of spousal aggregation does not appear useful in identifying unsuspected environmental causes of cancer in heterogeneous populations in urban areas of affluent Western countries. A cohort study would have to be much larger than this one to detect weak spousal concordance reliably.