Water Quality Trading (WQT) programs may offer farmers compensation to adopt Best Management Practices (BMPs). We conducted a survey of farmers in the Kentucky River watershed from 2011 to 2012. With respect to the five types of BMPs considered in the survey, about 20% of respondents did not indicate how much they will adopt. Missing responses are common for surveys on farming decisions. We compare three methods to handle the missing data: deleting the observations with missing value, mean imputation, and Multivariate Imputation by Chained Equation (MICE). Following these missing data treatments, we estimate the factors affecting how much farmers may engage in BMPs using Tobit or Poisson model. The results show that increasing the compensation for using BMPs is more likely to encourage farmers to adopt riparian buffers. In addition, land area, percentage of household income from farming, percentage of total household income reinvested back to farm, and current experience of BMPs will affect BMP adoption. The results obtained after using the MICE are more promising and reasonable than using the deletion or the mean imputation method. Implications are discussed for farmers’ BMP adoptions under WQT while missing observations are present.