Breastfeeding duration has been linked to maternal and child health. Current-status (or interval-censored) data offers a more reliable measure of duration and cessation of breastfeeding than approaches that rely on recall, due to high percentage of 'heaping'. The present study explores the determinants of breastfeeding termination within the first three years of life in Bangladesh using the Cox proportional hazards model. Data were drawn from two rounds of the Bangladesh Demographic and Health Surveys 2011 and 2014. Results showed that secondary or higher educated mothers (hazard ratio, [HR] 1.7), mothers from wealthier families (HR 1.4), and relatively over-nourished mothers (HR 1.2) were at higher risk of breastfeeding termination. Breastfeeding termination was also associated with contraceptive use, exposure to family planning programme in media and place of residence. Pooled data also suggests that early breastfeeding termination is increasing in Bangladesh. The findings have implications for policy makers in shaping public information campaigns on breastfeeding.