In general, the efficiency of microfinance-driven poverty-alleviation programs run by Government (GO) and Non-Government Organizations (NGOs) in developing countries such as Bangladesh is assessed by repayment rate, number of beneficiaries, area coverage, amount of loans disbursed, cost of operation, profitability and the financial sustainability of the projects. This thesis argues that these methods are quite restrictive since none of them reflect the perceptions of poor people concerning the efficiency of the projects. The development agencies (such as GOs and NGOs) have never been compared on the basis of service delivery efficiency [a processed based comparison], nor on their relative contribution to raising the living standards of the poor [an outcome based comparison]; both of which are crucial for poverty reduction. The main reason behind the absence of such comparisons is the unavailability of the appropriate parameters that could be used for such a comparison. -- For the process-based comparison this thesis develops and validates a two dimensional multi-item scale through construct, convergent, discriminant and nomological validity. The scale captures different aspects of efficiency in the delivery of services. These aspects are termed the 'credibility dimension' and the 'focus towards beneficiaries dimension' of the service delivery in poverty-alleviation programs. For the output-based comparison, this thesis also develops and validates a 'multidimensional poverty model' to compare the efficiency of GOs and NGOs in contributing to the economic, social, political and cultural elements in the lives of the poor. The methodology is based on 930 samples collected from 12 districts and 107 randomly chosen villages in Bangladesh during September-December 2009. -- The results from utilizing the scale items for comparative study show that in many important dimensions of service delivery, the performance of GOs is more efficient than that of NGOs. Gender discrimination also is evident in poverty-alleviation programs in Bangladesh. This study reveals that female beneficiaries are disadvantaged not only due to cultural or religious customs, but also because they receive less attention from the managers and field staff of both GOs and NGOs. -- The poverty model shows that as a whole GO agencies are more efficient in improving the welfare of poor beneficiaries compared to NGOs. While the comparative study shows that government agencies are more efficient in improving the economic wellbeing of the poor, NGOs are better at improving the social aspects of their lives.