Technical Abstract: The last outbreak of highly pathogenic (HP) avian influenza (AI) in North America involved an H7N3 AI virus (AIV) in Chile while the two previous outbreaks involved H5N2 HPAIVs. Both outbreaks began as mild respiratory disease and/or drops in egg production in poultry and abruptly changed to severe, systemic diseases that coincided with specific nucleotide changes in the hemagglutinin gene. A similar outbreak of H7N1 AI in Italy during 1999 began as low pathogenicity (LP) AI and abruptly changed to HPAI. Several laboratory models have been developed to derive HPAIVs from LPAIVs and study the mechanisms for this virulence shift. In our laboratory, we have studied 36 H5 and H7 LPAIVs for potential to become HPAIVs in a 14-day-embryonating-chicken-egg-adult-hen model system. Twenty-one H5 LPAIV isolates obtained from poultry and ratites were manipulated in a 14-day-embryo modeling system. Eight H5 AIV isolates changed from producing a few deaths (LP) to killing $75% of chickens in an intravenous pathogenicity test plus one derivative produced 100% cytopathic effect without exogenous trypsin in chicken embryo fibroblast cultures, but was not tested for in vivo pathogenicity. Three of the nine HP derivatives had the loss of an oligosaccharide site at amino acid residue 11 while one had an insertion of two basic amino acids at the proteolytic cleavage site. However, three derivatives lacked a consistent change in the hemagglutinin to explain the pathogenicity shift from parent to derivative. Two HPAIV derivatives were not sequenced. Fifteen H7 AIVs were tested in the model system and one isolate became HP, but this virus was essentially already a highly lethal virus following repeated IV pathotyping of the parent virus. Examination of genes other than the hemagglutinin may be necessary to fully understand and predict pathogenicity shifts of H5 and H7 AIVs.