BACKGROUND: In a previous community-based cohort study in Guinea-Bissau from 1996 to 1998, characterisation of rotavirus strains showed a high frequency of less common genotypes such as G8 and G9 and a high proportion of mixed infections. OBJECTIVES AND STUDY DESIGN: In the present study, we examined the prevalence of rotavirus genotypes among 81 hospitalised and 23 non-hospitalised Guinean children with rotavirus associated diarrhoea during the 2002 seasonal rotavirus outbreak. G- and P-types were determined in a two-step procedure using reverse transcription followed by a standard multiplex PCR. The multiplex PCR for G-types was furthermore supplemented with a single locus PCR including the MW8 primer for the G8-genotype. RESULTS: The dual infection G2/P[4]P[6] (24%) appeared to be the most frequent cause of rotavirus infections followed by G2P[4] (19%), G2P[6] (16%) and G8P[6] (13%). Overall 38% of the infections were mixed and 18% of the samples had the genotype G8. However, by subjecting all samples and not only the strains, which according to the standard multiplex PCR procedure were non-typeable, to a single locus G8-PCR, we found that the genotype G8 appeared in 62% of the infections, either as a single G-strain or in combination with other G-types, especially G2. Including these results, more than 63% of infections emerged as mixed. Neither genotype (including the presence of G8) nor the presence of mixed infections, seem to influence the severity of the rotavirus infection. CONCLUSION: We found a high frequency of mixed infections especially due to G8-genotypes, which might have implications for development of rotavirus vaccine candidates for use in Africa. Our results do not suggest that a single genotype is associated with severity, but the present study is based on a modest number of samples and results should be interpreted with caution.