Our study aims to contribute to this debate by providinglarger sample quantitative evidence on the organiza-tional factors that moderate the effectiveness of the useof performance measurement systems in public sectororganizations. We focus on two roles of the perfor-mance measurement system (incentive-oriented use andexploratory use), and on one key organizational character-istic (contractibility of performance). Building on economicand behavioural theories, we develop and test the modelthat is reflected in Fig. 1.We will argue in subsequent sections of this paperthat the cybernetic, incentive-oriented use of performancemeasures as advocated by NPM enhances performanceonly when contractibility is high. Relying on an incentive-oriented use when contractibility is low is likely to resultin an unbalanced attention to quantifiable results, ‘manag-ing measures rather than results’, and data manipulation(Marsden and Belfield, 2006; Newberry and Pallot, 2004).When contractibility is low, performance may benefitfrom an exploratory use of performance measures as thisprovides opportunities for double loop learning, mutualconsultation, and adaptation to emerging insights (cf.Burchell et al., 1980). Our central claim is that public sec-tor organizations that use their performance measurementsystems in ways that are consistent with the characteris-tics of their activities outperform organizations that fail toachieve such fit.We empirically test the model using survey data from101 public sector organizations. The results partially con-firm our theoretical expectations. Specifically, we findthat contractibility moderates the relationship between anincentive-oriented use of the performance measurementsystem and performance: while using the performancemeasurement system for incentive purposes negativelyinfluences organizational performance, this effect is lesssevere when contractibility is high. This finding suggeststhat NPM should allow for a more situation-dependentapproach to performance measurement. We also findthat an exploratory use of the performance measurementsystem tends to enhance performance in public sector orga-nizations. Contrary to our expectations, this positive effectappears to exist independent of the level of contractibility.These results suggest that public sector managers need not