This work determined scopes of arsenic(As)-contaminated groundwater using risk-based indicator classification approaches in blackfoot disease hyperendemic areas of southern Taiwan. Indicator kriging was first used to establish a conditional cumulative distribution function at each cell. Three approaches--the p-quantile estimate, the E-type estimate and the minimization of the expected loss--were then adopted to delimit contaminated regions for a regulated standard of As concentrations in groundwater. According to a risk assessment model established in our previous research, the standard was set to 250 microg/l for aquacultural use, corresponding to the 77.1th percentile of observed concentrations. Misclassification risks and uncertainty were examined for the classification approaches. The analyzed results reveal that contaminated areas are the largest using the 0.771-quantile estimate, whereas they are the smallest using the minimization of the expected loss. Proportions of credible polluted areas with low risks to false positives maintain a constant, 12.9-13.2%, for the classification approaches. To reduce a great impact on human health, As-polluted groundwater should be strictly prohibited to cultivate fish in credible polluted zones and monitored persistently in polluted zones with high risks to false positives.