The use of morphospecies as surrogates for taxonomic species has been proposed as an alternative to overcome the identification difficulties associated with many invertebrate studies, such as biodiversity surveys. Hymenoptera specimens were collected by beating and pitfall traps, and were separated into morphospecies by a non-specialist with no prior training, and later identified by an expert taxonomist. The number of Hymenoptera morphospecies and taxonomic species was 37 and 42, respectively, representing an underestimation error of 12%. Different families presented varying levels of difficulty, and although the species estimation provided by the use of morphospecies initially appeared to have a relatively minor error rate, this was actually an artefact. Splitting and lumping errors balanced each other out, wrongly suggesting that morphospecies were reasonable surrogates for taxonomic species in the Hymenoptera. The use of morphospecies should be adopted only for selected target groups, which have been assessed as reliable surrogates for taxonomic species beforehand, and some prior training to the non-specialist is likely to be of primary importance.