In April 1983, an influenza virus of low virulence appeared in chickens in Pennsylvania. Subsequently, in October 1983, the virus became virulent and caused high mortality in poultry. The causative agent has been identified as an influenza virus of the H5N2 serotype. The hemagglutinin is antigenically closely related to tern/South Africa/61 (H5N3) and the neuraminidase is similar to that from human H2N2 strains (e.g., A/Japan/305/57) and from some avian influenza virus strains (e.g., A/turkey/Mass/66 [H6N2]). Comparison of the genome RNAs of chicken/Penn with other influenza virus isolates by RNA-RNA hybridization indicated that all of the genes of this virus were closely related to those of various other influenza virus isolates from wild birds. Chickens infected with the virulent strain shed high concentrations of virus in their feces (10(7) 50% egg infective dose per g), and the virus was isolated from the albumin and yolk of eggs layed just before death. Virus was also isolated from house flies in chicken houses. Serological and virological studies showed that humans are not susceptible to infection with the virus, but can serve as short-term mechanical carriers. Analysis of the RNA of the viruses isolated in April and October by gel migration and RNA-RNA hybridization suggested that these strains were very closely related. Oligonucleotide mapping of the individual genes of virulent and avirulent strains showed a limited number of changes in the genome RNAs, but no consistent differences between the virulent and avirulent strains that could be correlated with pathogenicity were found. Polyacrylamide gel analysis of the early (avirulent) isolates demonstrated the presence of low-molecular-weight RNA bands which is indicative of defective-interfering particles. These RNAs were not present in the virulent isolates. Experimental infection of chickens with mixtures of the avirulent and virulent strains demonstrated that the avirulent virus interferes with the pathogenicity of the virulent virus. The results suggest that the original avirulent virus was probably derived from influenza viruses from wild birds and that the virulent strain was derived from the avirulent strain by selective adaptation rather than by recombination or the introduction of a new virus into the population. This adaptation may have involved the loss of defective RNAs, as well as mutations, and thus provides a possible model for a role of defective-interfering particles in nature.