Catch-per-unit effort (CPUE) has frequently been used as an index of abundance and more specifically to calibrate virtual population analysis (VPA). In multi-species fisheries, CPUE calculated from fishing trips targeting the species seems to be more effective for calibration than the classical ratio total landings/total effort. Target species are determined from an analysis of the composition of catches (landings) of each fishing trip, each trip being categorised as to whether it targets the given species or not. It is obvious that existing but not reported discards would affect these results. Classification of trips can be achieved on a single-species basis, each species being considered successively and each trip possibly being relevant for more than one species. Using a more general approach, classification can be achieved by métiers, each of these being determined by one or a group of target species and each trip being categorised into one and only one métier. This method of categorising trips is based on thresholds of target species contribution to the catch and on an overall explanatory level. For the main species of the French demersal fishery off the west coast of Scotland and in the Celtic Sea, CPUEs are calculated using different methods to define the trips used to calculate fishing effort and associated landings. Besides important differences in actual values, CPUEs may also differ in their trends depending on the definition of directed effort. Tuned VPA, carried out for some species, shows that large variations in population estimates, fishing mortality or short-term predictions could occur when using directed effort, while the catchability model fits the data better.