Objective: To identify predictors of recovery in children with uncomplicated severe acute malnutrition (SAM). Design: This is a secondary data analysis from an individual randomised controlled trial, where children with uncomplicated SAM were randomised to three feeding regimens, namely ready-to-use therapeutic food (RUTF) sourced from Compact India, locally prepared RUTF or augmented home-prepared foods, under two age strata (6-17 months and 18-59 months) for 16 weeks or until recovery. Three sets of predictors that could influence recovery, namely child, family and nutritional predictors, were analysed. Setting: Rural and urban slum areas of three states of India, namely Rajasthan, Delhi and Tamil Nadu. Participants: In total, 906 children (age: 6-59 months) were analysed to estimate the adjusted hazard ratio (AHR) using the Cox proportional hazard ratio model to identify various predictors. Results: Being a female child (AHR: 1 center dot 269 (1 center dot 016, 1 center dot 584)), better employment status of the child's father (AHR: 1 center dot 53 (1 center dot 197, 1 center dot 95)) and residence in a rental house (AHR: 1 center dot 485 (1 center dot 137, 1 center dot 94)) increased the chances of recovery. No hospitalisation (AHR: 1 center dot 778 (1 center dot 055, 2 center dot 997)), no fever, (AHR: 2 center dot 748 (2 center dot 161, 3 center dot 494)) and <= 2 episodes of diarrhoea (AHR: 1 center dot 579 (1 center dot 035, 2 center dot 412)) during the treatment phase; availability of community-based peer support to mothers for feeding (AHR: 1 center dot 61 (1 center dot 237, 2 center dot 097)) and a better weight-for-height Z-score (WHZ) at enrolment (AHR: 1 center dot 811 (1 center dot 297, 2 center dot 529)) predicted higher chances of recovery from SAM. Conclusion: The probability of recovery increases in children with better WHZ and with the initiation of treatment for acute illnesses to avoid hospitalisation, availability of peer support and better employment status of the father.