To estimate wolf (Canis lupus) kill rates from fine-scale movement patterns, we followed adult wolves in 3 territories of the Scandinavian wolf population using Global Positioning Systems (GPS) during the winters of 2001–2003. The resulting 6 datasets of 62–84 study days gave a total of 8,747 hourly GPS positions. We visited clusters of positions in the field on average 8.8 days after positioning and found moose (Alces alces) killed by wolves during the study period on 74 (8%) of the 953 clusters. The number of positions and visits to a cluster, their interaction, and the proportion of afternoon positions were significant fixed effects in mixed logistic-regression models predicting the probability of a cluster containing a wolf-killed moose. The models, however, displayed a poor goodness-of-fit and were not a suitable tool for estimating kill rates from positioning data alone. They might be used to reduce fieldwork by excluding unlikely clusters, although the reduction was not substantial. We discuss proximate factors (i.e., human disturbance and access to prey) as well as ultimate factors (i.e., social organization, intra-guild dominance, and litter size) as potential causes of the observed high temporal and spatial variation in prey-handling. For similar future kill-rate studies, we recommend increasing field efforts and shortening positioning intervals.