Matched serum and saliva samples were collected simultaneously from 124 subjects exposed during a hepatitis A virus (HAV) outbreak at a daycare center in Rio de Janeiro, Brazil. All samples were tested for IgM and total anti-HAV antibodies by enzyme immunoassay (EIA). HAV was detected by nested PCR in serum, saliva, and water samples employing primers for the VP1/2A region of the viral RNA; all positive products were then sequenced. The viral load of the matched samples was determined by real-time PCR using the TaqMan system. HAV-RNA was identified by nested PCR in 37.7% of the saliva samples, 29% of the serum samples, and one drinking water sample. The mean HAV viral load was similar in the serum and saliva specimens (10(3)  copies/ml). HAV genotypes IA and IB were detected in both specimen types, and the water sample isolate was classified as genotype IB, indicating the existence of more than one source of infection at the daycare center. In six infected patients, a different HAV subgenotype was found in their serum than in their saliva, and this unusual pattern of mixed HAV infection was investigated further by molecular cloning followed by nucleotide sequencing. All clones derived from the saliva samples belonged to subgenotype IB and shared 96.5-100% identity. However, clones derived from their corresponding serum sample belonged to subgenotype IA and shared 90.5-100% identity. This study showed the important role that non-invasive saliva samples can play in the molecular epidemiological analysis of a hepatitis A outbreak.