A case-control study was conducted in Melbourne, Australia of 88 consecutive males admitted for the surgical removal of a nonmelanocytic skin cancer (histologically confirmed basal cell carcinoma and squamous cell carcinoma) and of 88 male control patients admitted for small elective surgical procedures. In both cases and controls, previous diet, alcohol consumption, and smoking habit were investigated and serum beta-carotene and vitamin A levels were measured. A statistically significant inverse relationship was found between the risk of skin cancer and a high intake of fish (p = 0.05); vegetables in general (p < 0.001); beans, lentils, or peas (p < 0.001), carrots, silverbeet (Swiss chard), or pumpkin (p < 0.001); cruciferous vegetables (cabbage, brussel sprouts, or broccoli) (p < 0.001); and beta-carotene- and vitamin C-containing foods (p = 0.004). Cases had a lower mean serum level of beta-carotene (p < 0.001) and vitamin A (p = 0.02) than controls. The incidence of skin cancer in the study was inversely related to the level of serum beta-carotene (p < 0.0001). The correlation coefficient between dietary beta-carotene/vitamin C and serum beta-carotene was 0.22 (p = 0.04). Smoking and alcohol consumption showed no statistically significant association with the risk of nonmelanocytic skin cancer. The results were similar for both cell types. A high intake of vegetables including cruciferous vegetables, beta-carotene- and vitamin C-containing foods, and fish appears to be protective for nonmelanocytic skin cancer, and this deserves further study, as does the possible etiologic relevance of the low serum levels of beta-carotene and vitamin A.