OBJECTIVES: Mid-upper arm circumference (MUAC) and weight-for-height Z-score (WHZ) identify different populations of children with severe acute malnutrition (SAM) with only some degree of overlap. In an urban slum in Bangladesh, we conducted a prospective cohort study on children assessed as being severely malnourished by WHZ (<-3) but not by MUAC (>115 mm), to: 1. Assess their nutritional outcomes, and 2. Report on morbidity and mortality. METHODS: Children underwent 2-weekly prospective follow-up home visits for 3 months and their anthropometric evolution, morbidity and mortality were monitored. RESULTS: Of 158 children, 21 did not complete follow-up (six were lost to follow-up and 15 changed residence). Of the remaining 137 children, nine (7%) required admission to the nutrition programme because of: MUAC dropping to <115 mm (5/9 children), weight loss ≥ 10% (1/9 children) and severe medical complications (3/9 children, of whom one died). Of the remaining 128 children who completed follow-up, 91 (66%) improved in nutritional status while 37 (27%) maintained a WHZ of <-3. Cough was less frequent among those whose nutritional status improved. CONCLUSIONS: It seems acceptable to rely on MUAC as a single assessment tool for case finding and for admission of children with SAM to nutritional programmes.