Paddy productivity and variable factors efficiency were calculated based on a farm survey. Logit regression was employed to relate econometrically a set of farmer characteristics to indicators of pesticide exposure to identify types of health impairments that may be attributed to prolonged pesticide use. Then, the pesticides' negative effects on farmers' health were estimated by means of dose-response function. The empirical results indicated that the amount of pesticides applied was far higher than the optimal level for profit maximization. Insecticides influenced negatively and significantly farmers' health via the number of contacts rather than the total dose. Meanwhile, the higher the number of the doses and the number of applications of herbicides and fungicides, the bigger the health cost due to exposure. Since economic gains from input savings and a decrease in health cost outweighed productivity losses, a tax of 33.4 percent of pesticide price was proposed.