Microfinance is recognized as contributing both directly and indirectly to the Millennium Development Goals (MDGs). Using data from a survey of clients of a microfinance bank, Khushhali Bank, in 2005, the study revisited the survey data and found that despite the Bank's strict poverty-targeting program used in client selection and despite the survey's design to address the selectivity bias, the selectivity bias indeed still existed in the sampled households. Using the Propensity Score-Matching Methods (PSM) to address the selectivity bias, this study found that the lending program contributed significantly to income generation activities such as agricultural production and, in particular, animal raising (MDG 1). However, the impacts on other MDGs-education, health, female empowerment, and so forth-were of limited significance. This is due partly to the fact that 70% of the Bank's clients in the survey went through only one loan cycle, so the impacts on other MDGs are yet to be realized. Comparing the results to previous impact estimates done by Montgomery on the same dataset using OLS and Logit estimation, the PSM method yielded slightly different results. Although both studies recorded similar microfinance impacts on poverty, the degree of impact was less pronounced when the selectivity bias was addressed.