A distinct feature of neonatal rotavirus infection is the association of unusual strains that appear to be prevalent only in neonatal units and persist for long periods of time. The main aims of this study were to determine if rotavirus can be detected on environmental surfaces in the neonatal nursery and whether the infection occurs in mothers of infected and uninfected neonates. Thirty rotavirus positive neonates and an equal number of negative neonates were enrolled in this study. Stool samples from 15 mothers in each group and environmental swabs collected from the bed and surfaces around neonates were tested for rotavirus using single round and nested PCR for the VP6 gene. Rotavirus could be detected in environmental swabs using single round PCR for VP6 gene in 40% of neonates positive for rotavirus antigen by enzyme immunoassay (EIA) and 33.3% of EIA negative neonates. The detection rate was almost 100% using the nested VP6 PCR. Rotavirus was detected in maternal samples only if the nested VP6 PCR was used, with no significant difference between rates of rotavirus detection in maternal fecal samples of infected and uninfected neonates (p-0.4). Sequence analysis of nested VP6 amplicons from two environmental swabs revealed them to be closest in identity to G 10P[11 ], the most common genotype causing infections in neonates in this setting. Interestingly, sequences of amplicons from maternal stool samples did not cluster with G10P or other VP6 subgroup I strains but showed clustering with human strains of VP6 subgroup II.