In a recent paper dealing with wolf mortality in Italy (Lovari et al., 2007), the authors used a sample of 154 dead wolves incidentally found during a 11-year period to infer population parameters (e.g. sex and age ratios, mortality patterns, survivorship) and to provide a way to assess populationscale responses to conservation strategies. In our comment, we offer explanations as to why Lovari et al.’s (2007) paper has basic methodological flaws (e.g. inferences at the population level from an opportunistic sample of age at death and the use of static life-tables under violation of basic assumptions) that weaken the results, so that its conclusions are not warranted and should be cautiously interpreted. Generalizing from this specific case, we hereby argue that the use of opportunistic or convenience sampling (in this case, of dead animals) is not acceptable and should not be encouraged, especially if the results are used in population modelling in an applied perspective as these authors do, as many sources of bias can distort sample statistics from population parameters. Wildlife biologists working with endangered, low-density and elusive species such as large carnivores are constantly challenged to obtain robust and reliable population-scale datasets. These are needed to reliably assess the structure and dynamics of the populations, project their future trends and apply population models to conservation and management problems (Chapron et al., 2003). However, obtaining statistically, methodologically and biologically sound datasets at the population scale and for long time frames is not a trivial matter. Although researchers, especially dealing with endangered and threatened species, should strive to make the most out of any source of data, proper methodology and acknowledgment of potential sources of bias should be common practice ‘to get the basics right in wildlife field studies’ (Anderson, 2001) in order to provide ‘reliable knowledge’ (Romesburg, 1981). In a recent paper on wolf mortality (Lovari et al., 2007), the authors address the important topic of estimation of critical population parameters (i.e. population structure, reproduction, survivorship and mortality) for the wolfCanis lupus in central Italy, relating these parameters to current conservation strategies. By analysing a sample of incidentally found wolf carcasses, they conclude that ‘an extensive, routinely collection and analysis of wolf carcasses can be a relatively cheap but effective method to assess the state of a population, especially when data from the living population are missing’ (Lovari et al., 2007). We definitively concur with these authors that theirs is a worthwhile effort, as the lack of information at the population level undermines any rational approach to wolf management and conservation in Italy, as well in other European countries where the species is currently expanding its range (Boitani, 2003). However, we firmly believe that the use of these data (i.e. incidentally found carcasses) to infer population parameters and trends should not be encouraged, especially if the potential sources of biases are not adequately addressed. Although the authors recognize some limitations of their dataset, their concerns fall short and do not seem to have been stressed enough in their analyses. We believe that the major conclusions of the Lovari et al.’s paper are flawed by potential sources of bias that do not seem to have been adequately considered both in the results and discussion sections. In this letter, we explain why the basic nature of these flaws (e.g. inferences at the population level from an opportunistic sample of age at death, or the use of static life-tables under violation of basic assumptions) warrants some comments and criticism, in particular if the results are used in population modelling in an applied perspective (Chapron & Arlettaz, 2006). (1) In a truly representative sample of dead animals, each wolf dying in the population has the same probability of being reported, irrespective of its sex, age, social status, location, proximity to humans and cause of deaths. The sample of 154 dead wolves, incidentally found in central Italy by different observers (foresters, game wardens, etc.)