The 82 undergraduate students (42 male and 40 female) and 43 law students (34 male and 9 female) at the University of North Carolina heard either a male or a female witness in a taped reenactment of criminal trial testimony. The testimony was presented Subjects' in a 'fragmented' style, with brief answers by the witness to many questions by the lawyer, or in a 'narrative' style, with long answers to few questions. Consideration of adversary court norms and sex stereotypes led to the prediction that subjects would attribute favorable evaluation of the witness by the lawyer in the female witness-narrative style condition and unfavorable evaluation of the witness by the lawyer in the male witness-fragmented style condition. The prediction with respect to the female witness was confirmed only with law students; the prediction with respect to the male witness was confirmed only with undergraduates. Subjects< own evaluations of the witness showed the same pattern of effects. These results imply that the general network of power relations, norms, and stereotypes, and not the court setting per se, produces these interpretations of conversation styles. With regard to the issues in the social psychology of law, this experiment demonstrates that stylistic variation in testimony has substantial effects on perceptions of the court situation and on attributions about the individuals involved in a legal case. Findings also justify the interest of legal tacticians in questions of testimony style. Two tables and 19 references are included. (Author abstract modified)