The association between persistent diarrhoea and 'recent morbidity' defined as that occurring within the two-month period immediately preceding the onset of persistent diarrhoea was investigated in a population-based case-control study in rural North India. In two separate matched case-control analyses children with persistent diarrhoea (cases) were compared to population controls (five controls matched to each case) and acute diarrhoeal controls (three controls matched to each case). After correcting for possible confounding variables, comparing children with persistent diarrhoea and matched population controls, presence of a recent diarrhoeal illness was significantly associated with persistent diarrhoea with an odds ratio (OR) 2.6 (95%) confidence interval (CI): 1.1-7.1; p less than 0.05); during infancy this OR was 5.2 (95% CI: 1.0-31.9; p less than 0.01). Comparing children with persistent diarrhoea to matched acute diarrhoeal controls, presence of recent diarrhoeal illness was associated with an OR of 5.1 (95% CI: 1.3-20.3) in favour of the episode becoming persistent; in infants this OR was 10.4 (95% CI: 1.1-132.4; p less than 0.001).