It was hypothesized that an increase in the organizational value placed on convictions and improved evidence collection skills would result in fewer nonconvictions. Interventions included increased supervisory review of investigative reports and use of investigation and postarrest guides. Planned interventions involving feedback and decision and disposition reports were only partially implemented in the form of informal interim feedback reports. Data were collected in evidence, charging decision, and disposition in all felony arrests involving patrol officers in 5 precincts during 1984 and 1985: these included 609 control and 567 experimental pre- and posttest cases. While no experimental effect was found, the availability of several types of evidence increased for both samples during the posttest period. Both experimental and control cases showed a decrease in charge reductions during initial prosecutorial review and an increase in convictions on reduced charges at final disposition. Neither group showed changes in early dismissal rates or conviction rates, nor were there intergroup differences in dismissal rates or reasons for nonconviction. Factors contributing to the largely negative findings are discussed, and supplemental study information is appended. 4 figures, 42 tables, and 35 references. (Author abstract modified)