Spot urine samples were collected from the inhabitants of two rural communities in northwestern Bangladesh. We compared arsenic levels in the urine samples ([As](u); n = 346) with those in water from tube wells ([As](tw); range < 1-535 microg/L; n = 86) on an individual basis. The small variation of [As](u) within subjects and highly positive correlation with [As](tw) indicate that [As](u) is a useful indicator of exposure. Analyses of [As](u) showed that creatinine correction was necessary, that [As](u) only reflected recent exposure, and that there were substantial interindividual differences for a given [As](tw) level. To evaluate the toxic effects of arsenic exposure, we constructed a system for rating skin manifestations, which revealed distinct sex-related differences. Comparison of males and females in the same households confirmed that skin manifestations were more severe in the males, and in the males of one community a dose-response relationship between [As](u) and the degree of skin manifestation was evident. The results of this study indicate that [As](u) in spot urine samples can be used as an exposure indicator for As. They suggest that there might be sex-related, and perhaps community-related, differences in the relationship between [As](u) and skin manifestations, although several confounding factors, including sunlight exposure and smoking habits, might contribute to the observed sex difference. The existence of such differences should be further confirmed and examined in other populations to identify the subpopulations sensitive to chronic arsenic toxicity.