Geostatistical analysis and GIS-based spatial mapping have been widely used for risk assessment of environmental pollution. The objectives of this study were to: (1) investigate the spatial variability of pseudototal concentrations of Cd, Pb, and Zn; (2) estimate the degree of contamination on the basis of pollution indexes; and (3) combine geostatistical analysis with oral bioaccessibility to better assess the population's exposure to metals in smelter-impacted soils. Implications for human health risks were assessed by considering soil as a contaminant source, a release mechanism of contaminated soil to the hands, ingestion as an exposure route, and metal bioaccessibility. The bioaccessibility data in the gastric (G) and gastrointestinal (GI) phases were integrated into the standard hazard quotient-based risk assessment method. Using pollution indices showed that the entire area studied was highly polluted in terms of soil metal concentrations. However, the spatial pattern of health risk levels did not coincide with the spatial distribution of the degree of soil contamination. Introducing the bioaccessible fraction of metals from soils into the exposure calculations resulted in a substantial decrease in calculated risk (HI, hazard index) and provided a more realistic estimate of exposure to the three metals. For the highly exposed population, 46% of the soils studied provided an HI-G > 1.0 and 15% provided an HI-GI > 1.0, suggesting probable adverse health effects in children. The present study highlights the importance of conducting studies taking into account metal bioaccessible values in risk assessment.