Since the famine in 1974, the Bangladesh government and some national and international agencies have been providing food, or cash, or both, to poor households through the Social Safety Net programmes. I seek to estimate how much these programmes affect the well-being of poor households. Most previous studies have estimated impacts of these programmes on calorie consumption, simply computing the raw differential. However, both observed and unobserved characteristics bias this treatment effect. Using fuzzy regression discontinuity (RD) design, I control for these selection effects.