We use a large household level panel dataset collected from rural households in Bangladesh to examine the effects of microcredit program participation on household food security. We examine how microcredit affects different measures of food security; namely, household calorie consumption, dietary diversity indicators and anthropometric status of women of reproductive age (15-49 years) and children under the age of five. We find that microcredit program participation increases calorie consumption both at the intensive and extensive margins, but does not improve dietary diversity and only has mixed effects on the anthropometric measures. We also find that the effect of participation on food security may be non-linear in which participation initially has either no effect on food security or may actually worsen it, before improving it in the long run. Our results may explain why short-term evaluation of microcredit might not show any positive effects.