Using evolutionary theory to predict the dynamics of populations is one of the aims of evolutionary conservation. In endangered species, with geographic range extending over continuous areas, the predictive capacity of evolutionary‐based conservation measures greatly depends on the accurate identification of reproductive units. The endangered European eel (Anguilla anguilla) is a highly migratory fish species with declining population due to a steep recruitment collapse in the beginning of the 1980s. Despite punctual observations of genetic structure, the population is viewed as a single panmictic reproductive unit. To understand the possible origin of the detected structure in this species, we used a combination of mitochondrial and nuclear loci to indirectly evaluate the possible existence of cryptic demes. For that, 403 glass eels from three successive cohorts arriving at a single location were screened for phenotypic and genetic diversity, while controlling for possible geographic variation. Over the 3 years of sampling, we consistently identified three major matrilines which we hypothesized to represent demes. Interestingly, not only we found that population genetic models support the existence of those matriline‐driven demes over a completely panmictic mode of reproduction, but also we found evidence for asymmetric gene flow amongst those demes. We uphold the suggestion that the detection of demes related to those matrilines reflect a fragmented spawning ground, a conceptually plausible consequence of the low abundance that the European eel has been experiencing for three decades. Furthermore, we suggest that this cryptic organization may contribute to the maintenance of the adaptive potential of the species.