Although most conservationists claim to protect "species", the conservation unit actually and practically managed is the individual population. As resources are not unlimited, we need to focus on a restricted number of populations. But how can we select them? The Evolutionarily Significant Unit (ESU), first conceptualised by Ryder in 1986, may offer some answer. Several definitions have been proposed for the ESU, but all make reference to units "whose divergence can be measured or evaluated by putting differential emphasis on the role of evolutionary forces at varied temporal scales". Thus, an ESU might be fully identical with a "species", or a "species" could be composed of multiple ESUs. On the other hand, an ESU might comprise single/multiple populations exchanging a degree of gene flow, such as meta-populations. In an attempt to show strengths and weaknesses of ESU concepts, we present here, among several others, some case studies on the myrmecophilous butterflies of the genus Maculinea. In particular, we analyse the apparently everlasting debate about Maculinea alcon and M. rebeli, whose separation into separate species has been accepted by many authors, on mainly ecological criteria, but has not been fully supported by molecular analyses. We also discuss how the tight association with host ants may have driven selection for increasingly more strictly adapted Maculinea populations, arguably deserving specific taxonomic identity. Finally we discuss how current DNA analyses may fail to detect critical information on differences between taxa recently originated by the action of separate adaptive processes, which non-molecular studies can sometimes reveal. We conclude by discussing some current and often conflicting taxonomic trends, in their relationships with conservation policies.