BACKGROUND: The causes of oedematous vs non-oedematous childhood malnutrition (OM vs NOM) remain elusive. It is possible that inherited differences in handling oxidant stressors are a contributing factor. AIMS: To test for associations between polymorphisms in five genes and (i) risk of OM, a case-control study, and (ii) percentage cytotoxicity in peripheral blood mononuclear cells (PBMCs) exposed to hydrogen peroxide (H(2)O(2)), an in vitro cell challenge study. METHODS: Participants had been admitted previously for treatment of OM (cases, n = 74) or NOM (controls, n = 50), or were an independent set of healthy pregnant women (n = 47) who donated peripheral blood mononuclear cells. We tested for associations between genetic variation and outcome using single markers or a bivariate score constructed by counting numbers of deleterious alleles for each of 15 possible pairs of markers. RESULTS: In the case-control study there were no significant single-marker associations with OM. We did find that higher bivariate scores were associated with OM for the pair of NAD(P)H:quinone oxidoreductase 1 and catalase (odds ratio 2·00, 95% CI 1·05-3·82). In the cell challenge experiments, there were no significant associations with percentage cytotoxicity. CONCLUSIONS: Variation in this small set of genes seems unlikely to have a large impact on either risk of OM or cytotoxicity after H(2)O(2) exposure. The use of larger sample sizes to test the effects of a much larger set of genetic variants will be required in order to determine whether genetic variation contributes to the risk of OM. Such studies have potential for improving our understanding of causal pathways in OM.