One of the most intriguing questions studied recently in the field of computer-mediated communication (CMC) has been how communicators establish a 'social presence' in the absence of non-verbal cues which are relied upon heavily in face to face communication. One important area of social presence is the understanding of the gender of each of the participants in a conversation. Herring has speculated that, because of differential language cues, men and women can be identified in text-based messages (Herring, S.C. [1993]. Gender and democracy in computer-mediated communication. Electronic Journal of Communication [Online], 3(2). Available: http://www.cios.org/getfile\Herring-v3n293). The present study investigated the ability of readers of CMC messages to identify the gender of the author when messages were selected for language characteristics identified in previous studies (Savicki, V., Kelley, M. & Lingenfelter, D. [1996]. Gender, group composition and task type in small task groups using computer mediated-communication. Computers in Human Behavior, 12, 549-565.) as being associated with both group development and gender. Twenty messages from previous research were sorted into gender groups and into high and low communication style categories. Participants were asked their perception of the probable gender of the message author and their certainty of that judgment. Accuracy and certainty of judgments of gender showed significant differences between gender-communication style conditions. However, overall accuracy and certainty of judgments were not related. Neither was there a difference in accuracy or certainty of judgments between male and female judges. Finally, judges' accuracy followed gender stereotypes for messages sent by men, but were opposed to the stereotype for messages sent by women. (C) 1999 Elsevier Science Ltd. All rights reserved.