Human biomarkers were used to evaluate the lead (Pb) and arsenic (As) exposure of local people who lived in an agricultural area with intense agrochemical usage and who consumed groundwater. Although the heavy metals/metalloids in the groundwater were at low concentrations, they could cause adverse effects due to a high daily water intake rate over the long term. Biomarkers (hair, fingernails and urine) were collected from 100 subjects along with the local shallow groundwater and tap water, which is the treated deep groundwater, and investigated for the concentrations of As and Pb. Shallow groundwater had an average pH of 5.21 ± 1.90, ranging from 3.77 to 8.34, with average concentrations of As and Pb of 1.311 µg/L and 6.882 µg/L, respectively. Tap water had an average pH of 5.24 ± 1.63, ranging from 3.86 to 8.89, with the average concentrations of As and Pb of 0.77 µg/L and 0.004 µg/L, respectively. The levels of both As and Pb in the hair, fingernails and urine of shallow groundwater-consuming residents were greater than those in the hair, fingernails and urine of tap water-consuming residents. Interestingly, the As level in urine showed a linear relationship with the As concentration in groundwater (R2 = 0.91). The average water consumption rate was approximately two-fold higher than the standard; thus, its consumption posed a health risk even at the low As and Pb levels in the groundwater. The hazard index (HI) ranged from 0.01 to 16.34 (average of 1.20 ± 2.50), which was higher than the acceptable level. Finally, the concomitant factors for As and Pb in the urine, hair and nails from both binary logistic regression and odds ratio (OR) analysis indicated that groundwater consumption was the major concomitant risk factor. This study suggested that direct consumption of this groundwater should be avoided and that the groundwater should be treated, especially before consumption. In conclusion, urine is suggested to be a biomarker of daily exposure to As and Pb, while for long-term exposure to these metals, fingernails are suggested as a better biomarker than hair.