Police officers and probationer constables rated 40 sets of circumstances as to how far each would influence them to prosecute or not prosecute a motorist stopped for speeding. The distribution of responses about the ipsative means exhibited an asymmetry, consistent with the maximum salience bias towards a negative proportion of lle (Tuohy and Stradling, 1987). This asymmetry occurred in both directions, differentiating between a minority group for whom 'prosecute' was the negative pole and a majority for whom 'not prosecute' was negative. These groups did not differ significantly on overall mean rating, but did so on the primary underlying factor ('lack of deference'), where the former group was significantly more likely to prosecute. Significant differences were also found between probationer and experienced officers in their susceptibility to the bias, and in their distribution between the two groups. This paper reports on part of a large-scale investigation into discretionary practices in a major provincial police force in Britain. The substantive findings are detailed elsewhere (Roberts, Harper, Stradling and Quinn, 1988;Stradling and Harper, 1988); the present analysis, however, warrants separate treatment on a number of grounds. Firstly, in identifying distinct attitudinal schemas in a specific area of police work, the findings have potentially important implications for force training and operational administration. Secondly, in addition to what they suggest about the bases of discretionary judgment by police officers, the findings bear on a number of theoretical issues of some interest and complexity, notably the pervasive asymmetry of evaluative judgment (Adams-Webber, 1979) and its relationship to the mediation of affect (Tuohy, 1987). Methodologically, they also confirm and extend previous findings (Tuohy and Stradling, 1987), and show that judgemental asymmetry constitutes a useful technique whereby psychological differences may be identified as underlying actions in real-world situations.The relevant part of the investigation consisted of an instrument which was originally devised for use in a study of probationer constables (Harper, 1989). The