Children hospitalised with severe malnutrition have high mortality and readmission rates post-discharge. Current milk-based formulations target restoring ponderal growth but not the modification of gut barrier integrity or microbiome which increases the risk of gram-negative sepsis and poor outcomes. We propose that legume-based feeds rich in fermentable carbohydrates will promote better gut health and improve overall outcomes. We conducted an open-label phase II trial at Mbale and Soroti Regional Referral Hospitals, Uganda, involving 160 children aged 6 months to 5 years with severe malnutrition (mid-upper arm circumference (MUAC) < 11·5 cm and/or nutritional oedema). Children were randomised to a lactose-free, chickpea-enriched legume paste feed (LF) (n 80) v. WHO standard F75/F100 feeds (n 80). Co-primary outcomes were change in MUAC and mortality to day 90. Secondary outcomes included weight gain (> 5 g/kg/d), de novo development of diarrhoea, time to diarrhoea and oedema resolution. Day 90 MUAC increase was marginally lower in LF v. WHO arm (1·1 cm (interquartile range (IQR) 1·1) v. 1·4 cm (IQR 1·40), P = 0·09); day 90 mortality was similar (11/80 (13·8 %) v. 12/80 (15 %), respectively, OR 0·91 (95 % CI 0·40, 2·07), P = 0·83). There were no differences in any of the other secondary outcomes. Owing to initial poor palatability of the LF, ten children switched to WHO feeds. Per-protocol analysis indicated a trend to lower day 90 mortality and readmission rates in the LF (6/60 (10 %) and 2/60(3 %)) v. WHO feeds (12/71(17·5 %) and 4/71(6 %)). Further refinement of LF and clinical trials are warranted, given the poor outcomes in children with severe malnutrition.