Interest in virtual groups has focused on attribution biases due to the collocation or distribution of partners. No previous research examines self-attributions in virtual groups, yet self-attributions--the acknowledgment of personal responsibility or its deflection--potentially determines learning and improvement. This study reviews research on attributions in virtual groups and the effects of distance on members' proclivity to blame others or themselves. An experiment involved groups whose members were geographically collocated, distributed, or mixed, working over 2 weeks exclusively using asynchronous computer-mediated communication. Attributions for participants' own poor performance reflected a self-serving bias in completely distributed groups, whose members' eschewed personal responsibility and blamed their partners more than in collocated groups. Mixed groups' results help distinguish among competing theoretical perspectives. Moreover, an externally imposed observational goal mitigated attributional bias among distributed members by raising awareness of the sociotechnical effects of communication medium among those for whom the goal was successfully induced.