Six scat-analysis methods were compared and tested for differential assessment of a wolf Canis lupus diet in the Northern Apennine Mountains, Italy. A sample of 217 wolf scats was analysed using standardised laboratory techniques, and the recovered undigested remains were quantified according to the following diet measurements: frequency of occurrence, dry weight (estimated and measured), relative volume, and biomass ingested (two methods). With the exception of one of the biomass methods, there was no significant disagreement between the procedures examined. However, some discrepancies between rankings from different methods indicated the sources of bias that should be accounted for to avoid misleading conclusions. Frequency data can be corrected to reduce some of the associated forms of bias, whereas rankings by weight and volume appear affected by the structure of undigested remains. Although to different extents, all the methods which rank food items according to direct measures of the undigested remains, i.e. by frequency, weight, and volume, suffer from the surface to volume ratio bias of varying prey sizes. Linear-regression biomass models correct for the surface/volume bias, but there are some drawbacks when applying them, and they are limited to mammalian prey. Applicability of the biomass models should be evaluated on the basis of diet composition and prey sizes, and results carefully interpreted in concert with other field-collected information. Interpretation of scat-analysis data in order to assess the diet of wolves, as well as of other carnivores, would be greatly enhanced by comparing results obtained with two or more methods.