We implement a mixed method approach that combines a randomized controlled trial and qualitative data collection to assess whether, and if so how, behavioural change can be sustained. We do so in the context of Pakistan’s national sanitation strategy to combat open defecation, Community-Led Total Sanitation. Our findings demonstrate that continued follow-up activities, that build on the original intervention, reduced reversal to unsafe sanitation, but only where initial conditions are unfavourable —i.e. poor public infrastructure and low-quality sanitation facilities. Promotion efforts are hence best targeted towards those that face larger difficulties in constructing and maintaining high quality sanitation.