This study is an investigation into the influence of activity on otter (Lutra lutra) spraint production and composition, based on feeding trials involving captive, tame otters. Trials were of two types: ‘active’ where animals were encouraged to run, swim and dive for at least 33% of the time during the first six hours after each meal, and ‘inactive’ where activity varied between zero and 18% of this period. Trial types were balanced for number of fish, meal volume and fish size. Activity was found to have a large effect on spraint production and composition. For active trials, median minimum gut transit time was 67 minutes and median number of spraints per trial was 23. Corresponding figures for inactive trials were 170 minutes and 14, respectively. Of ingested salmonid (Salmo spp.) atlas vertebrae, 60% were recovered in spraints after active trials, while only 32% were recovered after inactive ones. Analyses of these effects showed that our previous models for estimating prey size distribution from bones recovered in spraints gave a better fit for inactive than for active trials. As otters in the wild are likely to spraint mostly during periods of activity, especially foraging, further feeding trials were undertaken to refine an earlier model better to mimic the field situation. Trials, involving active otters, were undertaken to: (i) produce in vivo bone to fish length equations for salmonid atlases and eel (Anguilla anguilla) thoracic vertebrae; and (ii) model the size-related differential recovery of these bones. These refined equations are more soundly based to provide estimates of prey size distribution than the earlier ones and their use in future studies of size-selection is recommended. The present study also highlights the potential limitations of extrapolating from the captive situation to the wild.