Little empirical work exists measuring if interagency collaborations delivering public services produce better outcomes, and none looking inside the black box at collaboration management practices. We examine whether there are collaboration management practices associated with improved performance of Crime and Disorder Reduction Partnerships, a crossagency collaboration in England and Wales. These exist in every local authority in England and Wales, so there are enough of them to permit quantitative analysis. And their aim is crime reduction, and crime data over time are available, allowing actual results (rather than perceptions or self-reports) to be analyzed longitudinally. We find that there are management practices associated with greater success at reducing crime, mostly exhibited through interaction effects such that the practice in question is effective in some circumstances but not others. Our findings support the arguments of those arguing that effective management of collaborations is associated with tools for managing any organization, not ones unique to managing collaborations: if you want to be a good collaboration manager, you should be a good manager, period.