The authors conducted a case-comparison chart-review study on 498 patients with histologically confirmed salivary gland carcinoma and 487 randomly selected patients registered at The University of Texas M.D. Anderson Hospital and Tumor Institute at Houston from 1960 through 1981. The study was undertaken to identify potential risk factors for salivary gland carcinoma. Univariate analysis of the data indicated a relationship with prior radiation therapy (odds ratio estimate of 6.17 with 95% confidence limits of 3.11-12.22) and with previous primary cancer (odds ratio estimate 4.81 with 95% confidence limits of 2.70-8.55). For agricultural occupations, the odds ratio was 1.62 with confidence limits of 1.05 to 2.49. The excess for previous primary cancer was largely due to previous skin cancer, especially in men, for whom the odds ratio was 13.7 (4.17-44.97). Multivariate analysis of these data, using the logit model, confirmed the association with both previous primary cancer and previous radiation. The rationale for a possible relationship between cutaneous neoplasms and salivary carcinoma is explained in embryologic and histogenetic terms.