In temperate and arid climate zones many streams and rivers flow intermittently, seasonally contracting to a sequence of isolated pools or waterholes over the dry period, before reconnecting in the wetter parts of the year. This seasonal drying process is central to our understanding of the population dynamics of aquatic organisms such as fish and invertebrates in these systems. However, there is a dearth of empirical data on the temporal dynamics of such populations. We describe a spatially explicit individual-based model (SEIBM) of fish population dynamics in such systems, which we use to explore the long-term population viability of the carp gudgeon Hypseleotris spp. in a lowland stream in southeastern Australia. We explicitly consider the impacts of interannual variability in stream flow, for example, due to drought, on habitat availability and hence population persistence. Our results support observations that these populations are naturally highly variable, with simulated fish population sizes typically varying over four orders of magnitude within a 50-year simulation run. The most sensitive parameters in the model relate to the amount of water (habitat) in the system: annual rainfall, seepage loss from the pools, and the carrying capacity (number of individuals per cubic meter) of the pools as they dry down. It seems likely that temporal source sink dynamics allow the fish populations to persist in these systems, with good years (high rainfall and brief cease-to-flow [CTF] periods) buffering against periods of drought. In dry years during which the stream may contract to very low numbers of pools, each of these persistent pools becomes crucial for the persistence of the population in the system. Climate change projections for this area suggest decreases in rainfall and increased incidence of drought; under these environmental conditions the long-term persistence of these fish populations is uncertain.