Four-hundred children below 3 years of age with acute watery diarrhoea were studied over a 12-month period for detection of rotavirus (RV) antigen in the stool samples. Detection of RV antigen in the cases has been compared with 175 age-matched non-diarrhoea controls. Overall positivity for detection of RV antigen was 6 per cent which is much lower than earlier studies. Even during the peak season the RV antigen detection was positive only in 13 per cent cases. However, RV antigen detection was significantly higher in cases than in the controls (P < 0.05). No significant correlation was noticed between the detection of RV antigen in the stool and environmental temperature, rainfall, and relative humidity except for higher percentage of RV antigen detection between February and April. Clinical profile of cases with RV antigen-positive stools has been compared with RV antigen-negative controls. Preceding history of vomiting and respiratory symptoms were noticed in higher percentage of RV antigen-positive cases, but no reliable 'clinical predictor' could be identified to differentiate RV diarrhoea from non-RV diarrhoea. Presence or absence of fecal polymorphonuclear leucocytes do not seem to rule out or suggest RV diarrhoea.