1. When strains of influenza A virus which have been isolated in chick embryos are introduced into the mouse lung, the virus multiplies readily and achieves initially a titer which is as high as is even obtained, even after repeated passage. The high initial titer of virus may be unaccompanied by any lethal or visible pathogenic effects; but with four or five mouse passages the agent becomes lethal in high titer and causes extensive pulmonary consolidation, though its capacity to multiply in the lung has not increased. In one example the adaptation to mouse lung was accompanied by increasing capacity to agglutinate guinea pig red cells without a corresponding increase in agglutinating power for chicken cells. Influenza B virus, in preliminary tests, did not behave in a similar fashion. 2. The adaptation of influenza A virus to mice is accompanied by changes in antigenic pattern, as detected by cross-tests with the agglutination inhibition method. Two strains, initially similar, with passage, changed in pattern along divergent paths so that they became not only unlike the parent strains but unlike each other. This finding has important implications for the interpretation of the strain difference problem in human influenza.