This study investigated the effects of witness discussion on the accuracy of recall and misidentifications in eyewitness memory. Dyadic groups who discussed the crime and dyads who made joint memorial decisions about the incident were more accurate in recall than were dyads who discussed noncriminal matters, and a no-discussion group. Collaborative dyads giving joint decisions also were significantly more accurate than were the other groups in rejection of the culprit-absent lineup. The sequential lineup method proved superior to the simultaneous method, regardless of whether or not witnesses discussed the crime prior to identification. It was concluded that discussion and joint decision making by 2 eyewitnesses can improve both eyewitness recall and minimize false identifications in some circumstances.