Freshwater systems offer important opportunities to investigate the consequences of intrinsic biological and extrinsic environmental factors on the distribution of genetic variation, and hence population genetic structure. Drainages serve to isolate populations and so preserve historical imprints of population processes. Nevertheless, dispersal between and within drainages is important if the biology of the species confers a good dispersal capability. Knowledge of the population genetic structure or phylogeographic patterns of Australia's freshwater fish fauna is generally depauperate, and the present study aimed to increase this knowledge by investigating patterns of genetic diversity in three Australian species of freshwater fish. I was interested in the relative importance of dispersal capability, the hierarchical nature of stream structure and the consequences of earth history events on patterns of genetic diversity among populations. I examined three species from three families of Australian freshwater fish, Pseudomugil signifer (Pseudomugilidae), Craterocephalus stercusmuscarum (Atherinidae) and Hypseleotris compressa (Gobiidae). These species are abundant, have wide overlapping distributions and qualitatively different dispersal capabilities. I was interested in attempting to unravel how the biological, environmental and historical factors had served to influence the patterns and extent of genetic diversity within each species, thereby inferring some of the important evolutionary processes which have affected Australia's freshwater fauna. I used allozyme and 500-650bp sequences from the ATPase6 mitochondrial DNA (mtDNA) gene to quantify the patterns of genetic variation at several hierarchical levels: within populations, among populations within drainages and among drainages. I collected fish at several spatial scales, from species wide to multiple samples within drainages; samples were collected from the Northern Territory, Queensland and New South Wales. The species with the highest potential for dispersal, H. compressa, exhibited the lowest levels of genetic differentiation as measured at several allozyme loci (H. compressa: FST=0.014; P. signifer FST=0.58; C. stercusmuscarum FST=0.74). Populations of H. compressa also had low levels of mtDNA differentiation, with many recently derived haplotypes which were widespread along the coast of Queensland. This suggested either considerable gene flow occurs or recent demographic change in the populations sampled. As there was no relationship between geographic distance and genetic differentiation, the populations appeared to be out of genetic drift - gene flow equilibrium, assuming the two-dimensional stepping stone model of gene flow. Estimating contemporary gene flow was thus difficult. It was apparent that there has been a recent population expansion and / or contraction of H. compressa populations. It was concluded that there has been considerably more connectivity among populations of H. compressa in the recent past than either of the other study species. Populations of P. signifer showed considerable genetic subdivision at different hierarchical levels throughout the sampled range, indicating gene flow was restricted, especially between separate drainages. Two widely divergent regional groups which had high ATPase6 sequence divergence and approximately concordant patterns at allozyme loci were identified. Interestingly, the groups mirrored previous taxonomic designations. There was also significant subdivision among drainages within regional groups. For example, the adjacent Mulgrave-Russell and Johnstone drainages had individuals with haplotypes that were reciprocally monophyletic and had large allozyme frequency differences. This allowed me to examine the patterns of genetic differentiation among populations within drainages of two essentially independent, but geographically close systems. There was as much allozyme differentiation among populations within subcatchments as there was between subcatchments within drainages, and significant isolation by distance among all populations sampled within a drainage. This suggested that the estuarine confluence between subcatchments was not a barrier to P. signifer, but that distance was an important component in the determination of the distribution of genetic diversity within drainages in P. signifer. There were three main areas of investigation for C. stercusmuscarum: comparing upland and lowland streams of the drainages in north Queensland, investigating the consequences of eustasy on coastal margin populations and examining the intriguing distribution of the two putative sub species, C. s. stercusmuscarum and C. s. fulvus in south east Queensland. First, as populations in upland areas of east coast flowing rivers are above large discontinuities in the river profile, their occurrence is presumably the result of gene flow to and / or from lowland areas, or the result of invasions via the diversion of western flowing rivers. Concordant patterns at both genetic markers revealed that the latter possibility was the most likely, with fixed allozyme differences between upland and lowland populations, and large mtDNA sequence divergence. Indeed, it appeared that there may have been two independent invasions into the upland areas of rivers in North Queensland. Second, lowland east coast populations also had large, although not as pronounced, levels of population subdivision. Lack of isolation by distance, but with a concomitant high level of genetic differentiation among many comparisons, was consistent with a scenario of many small, isolated subpopulations over the range. Interestingly, widespread populations in central Queensland coastal populations (drainages which receive the lowest rainfall) were relatively genetically similar. This was consistent with the widest part of the continental shelf which at periods of lower sea level apparently formed a large interconnected drainage, illustrating the effect of eustatic changes on populations inhabiting a continental margin. Third, putative C. s. fulvus in lowland coastal Queensland drainages were genetically more similar to a population of C. s. fulvus collected from a tributary of the Murray-Darling (western flowing) than they were to adjacent putative C. s. stercusmuscarum. This implied that populations in south east Queensland, north to approximately the Burnett River, appeared to be derived from western flowing streams, and not via dispersal from other lowland east coast populations. Determining the relative importance of intrinsic and extrinsic factors to the development of population genetic structure is a difficult task. The present study demonstrated that the species with the highest dispersal potential had the lowest levels of genetic differentiation, waterfalls can limit gene flow, eustasy acts to join and separate populations leading to complex genetic patterns and that drainage rearrangements are important in determining the distribution of genetic diversity of populations now inhabiting isolated drainages. A difficulty with generalising about population genetic structure in obligate freshwater animals is the unique history of not only each drainage, but also the streams within that drainage and the idiosyncratic biological dynamics of the populations inhabiting those drainages.