This study investigated the effects of police-specific, crisis intervention, communication training upon two dependent variables: observable police interviewing behavior, and self-reported communication apprehension. The interview measurement, designed for the study, was comprised of three elements: introduction behavior, 7 helper behaviors, and information gathering and giving. A third dependent variable, a predictive index, was used to determine whether to train entire departments, or volunteer specialists. Police officers from a single department, matched on three relevant variables, served as trainee/subjects in two conditions. Twenty-three experimental trainees participated in a 32-hour program developed for the study, while 23 control trainees received 12 hours of pseudo-training. Each group contained equal numbers of patrolmen, detectives, and sergeants. Two lieutenants were placed in the experimental group, but trained in two separate units. This was the sole departmental restriction upon the design. Subjects were trained in alternating units of 11 or 12, and the investigator served as sole trainer. Experimental training took the form of active practice with exercises relating to the skills deemed essential for success during family crisis mediation. Major training elements were: increasingly complex role play with police-specific scenarios; systematic feedback from peers, trainer and video confrontation, and modeling. The design was one of multi, repeated measures. A MANOVA F test was performed for overall examination, while ANOVA F tests determined the specific contribution made by each measure. Analysis examined two distinct trial contrasts, baseline versus posttest, and posttest versus follow-up. The MANOVA F test revealed significant differences for group main effect (F(3,42) = 6.11, p < .01), trial main effects, trial 1 versus trial 2, (F(3,42) = 3.08, p < .05), and the groups by trials interaction, trial 1 versus trial 2 (F(3,42) = 3.76, p < .05). The contrast of trial 2 versus trial 3 was nonsignificant in all cases indicating that change scores were maintained, and that a practice effect was not operative. The three measures contributed unequally to the significance with the interview test showing the largest contribution to change, followed by the communication apprehension test. The predictive index failed to function as such. Implications of the findings are discussed, as are the results of trainees1 evaluations of training. Subject and police department factors, unique to this study, are also noted. Suggestions for future research are detailed.