The research used a behavior observation procedure to record 40,661 behaviors during 530.75 hours on patrol. It included four male and four female police officers described by peers and supervisors as skilled at conflict resolution and eight male officers not so described. A hand-held data logger recorded all behaviors displayed by the officer being observed, the officer's partner, and the citizen. The logger was programmed to record the duration of each behavior and the sequence in which the behavior occurred. Police behavior categories included giving information, seeking information, controlling statements, rejecting, threats, supportive statements, and physical actions. Citizen actions included giving information, seeking information, self-defensive statements, refusals to comply, verbal abuse, and physical abuse. Results revealed complex relationships and differences between male and female police officers. Officers described as skilled were more active in an interaction than the average officers. Physical acts by officers were preceded by declining information exchange. The citizen had limited influence on subsequent police behavior during problematic encounters, although the citizen's refusal to respond was associated with an increase in the chance that the officer would respond physically. Finally, a physical act by a citizen was generally preceded by a physical act by a police officer. Figure, tables, footnotes, appended methodological information and additional results, and 34 references