Context. Concerns about conservation and ethics in amphibian research have been raised recently; therefore, non-lethal methods should always be preferred to investigate food habits in wild populations. However, there are no studies that compared the data obtained by different non-lethal dietary methods in amphibians.

Aims. We compared the dietary habits obtained from stomach flushing, a method validated against stomach dissection, and faecal analysis on the same 31 individuals of the European plethodontid Speleomantes strinatii, a completely terrestrial salamander.

Methods. After being stomach-flushed in the field, salamanders were kept in the laboratory at constant humidity and temperature to obtain faecal samples. Analyses comprised diversity and niche overlap indexes, as well as permutation, repeated-measures tests and graphical methods.

Key results. Niche overlap between the two samples was low (O_jk = 0.58) and prey diversity was significantly (P = 0.001) higher in stomach contents. There were also differences in the abundance of fly larvae, springtails and ants and the interpretation of the population trophic strategy varied according to the sampling method used.

Key conclusions. Stomach flushing and faecal analysis gave different information, because, apparently, a differential prey degradation occurred. During the digestive process, ants became dominant, whereas springtails and fly larvae became under-represented in faecal samples. Therefore, on the basis of faecal analysis, diet diversity was underestimated and many individuals improperly appeared as ant-specialist feeders.

Implications. In terrestrial salamanders, results from stomach flushing and faecal analysis should not be compared among studies, populations or species. Moreover, stomach flushing should always be preferred when assessing the trophic strategy and the role in food webs of salamanders.