Human respiratory syncytial virus (HRSV) is classified into two major groups, A and B, each of which contains multiple variants. To characterize the molecular epidemiology of HRSV strains over time, sequencing studies of a variable region of the attachment protein gene from a single community in the United States during 5 successive years were performed. Phylogenetic analysis revealed distinct clades (genotypes) that were further classified in subtypes based on > or = 96% nucleotide similarity. Five genotypes and 22 subtypes among 123 group A HRSV isolates, and four distinct genotypes and six subtypes among 81 group B HRSV isolates were identified. One to two genotypes or subtypes accounted for > or = 50% of isolates from a given year. A shift in the predominant genotype or subtype occurred each year such that no genotype or subtype predominated for more than 1 of the 5 study years. The consistency in the displacement of the predominant strain suggests that a shift, even within the same group, is advantageous to the virus. It was hypothesized that the 'novel' strain is better able to evade previously induced immunity in the population and consequently either circulates more efficiently or is more pathogenic. The yearly shift in HRSV strains may contribute to the ability of HRSV to consistently cause yearly outbreaks of HRSV disease. These results also suggest that isolates may need to be characterized as to both group and genotype to fully understand protective immunity after natural infection and efficacy studies of candidate vaccines.