Although microcredit is considered the main vehicle for increasing the income of the poor and alleviating poverty in Bangladesh, it is now well recognised that more than this is needed to reach the ultra poor in rural areas. Consequently, almost half of the Bangladesh population is in some way linked to non-governmental organizations' development programmes, at the centre of which is poverty alleviation. The study reported here developed a comprehensive cross-sectional intervention to examine whether the income of those who participated in the training programmes of non-governmental organizations was related to that participation. The study was carried out in three leading organizations in three districts of Bangladesh using a quantitative associational research design. Three hundred responses were analysed using multiple regression analysis to elicit information from ultra poor clients who had received training. The results suggest that the post-training income level of the trainees was negatively rather than positively related to the fact of their participation in training, the opposite of what human capital theory would lead us to expect.