In cave animals, different levels of gene flow among populations appear to be related to the degree of cave dependence of each organism (Caccone, 1985). In general, high degrees of cave dependence entail a reduced ability of organisms to survive in surface environments, which in turn strongly influences their dispersal ability. In nature, constraints for dispersal can be geographical and ecological.Sometimes barriers can be represented by fluvial barriers or drainage divides (Barr, 1985); in other cases, a non-continuous "superficial underground compartment" (Juberthie, 1980) may represent an ecological break in the routes for dispersal (Culver, 1982). The existence of morphological variation among cave populations is often interpreted as reflecting a reduction in gene flow across dispersal barriers. Sometimes, morphological and genetic data are congruent in indicating similar geographic variation patterns (Kane et al. , 1992). In other cases, these different sets of characters clash, revealing independent trends (Allegrucci et al., 1987). This fact may reflect the different weight of selective pressures: there is evidence that variation in these characters is related to the distribution of ecological factors, such as temperature (De Matthaeis et al., 1985; Sbordoni et al., 1991). On the other hand, local populations of cave-dwelling and endogean arthropods may be highly differentiated genetically without exhibiting corresponding morphological differentiation (Laing et al., 1976; Delay et al., 1980; Cobolli Sbordoni et al., 1980; Sbordoni, 1982). Thus variation in morpho1ogical features by itself is not a good predictor of the levels of gene flow among a group of cave populations. In these circumstances we studied a woodlouse, Androniscus dentiger (Isopoda, Oniscidea), inhabiting humid edaphic environments and both natural and artificial caves. In A. dentiger the occurrence of an appreciable geographical variation among populations is not revealed by any of the morphological characters examined so far. This species has a wide distribution, which may suggest the existence of gene flow among populations, especially between neighbours. To test this hypothesis, allozyme electrophoresis was carried out to investigate the genetic structure of 28 populations at 19 gene loci. Fst values indicate a high level of differentiation. Allozyme data were used to estimate levels of gene flow among populations. The methodological approach and the biologica! significance of the results obtained are discussed.