Fishing affects fish populations through direct and indirect effects. It can change size structures and/or modify population mean weights. Reference values are thus needed to assess the status of populations in exploited ecosystems. These reference values can either be set by a historical approach, i.e. using information from before the onset of exploitation or overexploitation, or by a simulation approach. Using a model based on life-history parameters, we predicted population mean weights and length structures at equilibrium (in the absence of fishing and for fishing equal to different fishing mortalities) which we compared with contemporary data collected during scientific surveys in the Bay of Biscay. Contemporary mean weights were 88% to 30% smaller than expected for unexploited populations for 10 out of the selected 18 demersal species. Part of this difference might be explained by the survey not covering all age classes in the population, as demonstrated for Merluccius merluccius. We found that species with larger asymptotic length and slower growth were generally more impacted by fishing than smaller, faster growing species. Assuming that species specific life-history traits are well documented and/or easily measurable, the simulation approach can provide a useful tool for setting indicator reference levels for mean weight and size structures.