Attribution researchers have considered the dispositional bias to be pervasive even though it has never been examined as a function of group discussion. Group discussion is a common context for attributions (e.g., jury deliberations) and there are reasons to expect such discussions to affect the bias. The current experiment examined this issue within an attitude attribution paradigm. Because group discussion necessitates a time delay between presentation of information and the attribution dependent measures, subjects in three control conditions were asked to estimate an essayist's attitudes after little deliberation, much deliberation, or after merely preparing for a discussion. The 472 male and female participants were assigned randomly to one of 16 conditions in a 2 (pro vs. con essay) × 2 (choice vs. no choice) × 4 (immediate judgment vs. delayed judgment vs. delayed plus anticipated discussion judgment vs. group discussion judgment) between‐subjects factoral design. Results indicated that group discussion significantly attenuated the dispositional bias but neither prolonged deliberation nor anticipated discussion had such effects. The fact that actual discussion was required to eliminate the dispositional bias, rather than the mere anticipation of such discussion, suggests that there is something about the sharing of arguments that helps attributors to avoid the dispositional bias. Analyses of the discussion transcripts, however, failed to reveal the content of the arguments that produced the attenuation effect.