The spawning population of European eel (Anguilla anguilla L.) has been considered panmictic on the basis of genetic markers and morphometric studies. This hypothesis was tested by screening glass eel from five locations (Ireland, Italy, Morocco, Sweden and U.K.), belonging to two cohorts at the cytochrome b (cyt b) locus (392 bp) of the mitochondrion and at five nuclear microsatellite loci. Seventeen cyt b haplotypes were detected, of which ten were singletons; the most common haplotype occurred in 47% of all fish. Haplotype number increased significantly with latitude. Phylogeographical structure based on the cytoplasmic marker was weak (F-ST = 0.014) and non-significant. Close similarity was revealed between British and Irish glass eel populations, and weak differentiation among the British/lrish, Atlantic Moroccan, Italian and Swedish Baltic populations, respectively. No hierarchical genetic structure was obvious. Levels of genetic variation detected with five microsatellites were much higher levels than found with allozymes in previous studies (mean number of alleles per locus = 11.1; mean expected heterozygosity = 0.68). Overall among-population microsatellite variance was low but significant (F-ST = 0.004), and caused by the linked microsatellite loci Aan03 and Aan04. The Hardy-Weinberg-Castle equilibrium and the absence of gametic disequilibria at these loci in the Moroccan population might point to its genetic isolation, although the impact of just two out of five loci is puzzling. Given the weak differentiation typical for marine species and the limitations of our data, the results should be interpreted with caution. However, combined with recent evidence from a related study, the paradigm that the European eel constitutes a panmictic population becomes difficult to maintain.