Considerable evidence has been found to support the hypothesis that actors' and observers' attributions diverge. It is also becoming clear that this divergence is not the result of a single process, but is the result of factors which co-occur and differentiate actors from observers. Observers' beliefs about their similarity to actors were manipulated in two studies to see whether attributional differences between actors and observers covary with informational differences. In the first study, 90 male undergraduates served either as actors, similar observers, or dissimilar observers. Actors either succeeded or failed in changing another's attitude. Significant actor-observer effects occurred on most attribution measures: open-ended internality, actor's motivation, task difficulty, and luck. Similar observers' attributions almost always fell between actors' and dissimilar observers' attributions. In the second study, 175 observers rated a fictional other's personality as less dependent "on the situation" when s/he was more dissimilar. The results support an informational, rather than a perspective or motivational explanation for the actor-observer divergence. (NRB)