The evidence linking arsenic in drinking water with increased urinary cancer risk comes from populations in relatively high exposure areas (>100 microg/L), whereas studies from lower exposure areas (<100 microg/L) reported inconsistent results. A previous study conducted in northeastern Taiwan, where residents were exposed to relatively lower concentrations, reported increased risk of urinary cancer in a dose-response way. Using the same cohort with longer follow-up, we conducted analysis to elucidate the relationship between ingested arsenic and urinary cancer in lower exposure groups and assessed the influence of duration, recency, and latency of drinking arsenic-containing well water. A total of 8,086 residents from northeastern Taiwan were followed for 12 years. Incident urinary cancer was ascertained through linkage with the national cancer registry. All analysis was done by Cox proportional hazards regression models. There were 45 incidences of urinary cancer and a monotonic increased risk of urinary cancer was found with increasing arsenic concentration (P < 0.001). For the highly exposed (>100 microg/L), the relative risks (RR) were >5-fold, whereas the risk was elevated but not significant for low exposure (<100 microg/L). Relative to the arsenic concentration <10 microg/L, those who drank well water with higher concentration from birth [RR, 3.69; 95% confidence interval (95% CI), 1.31-10.4], still drank at enrollment (RR, 3.50; 95% CI, 1.33-9.22), and drank for >50 years (RR, 4.12; 95% CI, 1.48-11.5) had a significantly increased risk of urinary cancer. When restricted to urothelial carcinoma, all risk estimates including concentration and characteristics of well water consumption were higher.