Most studies of community development in insular systems have investigated the colonization dynamics of only a portion of the biotic community using islands that have the benefit of prior biotic modification. Two experiments assessed the predictive ability of equilibrium island theory with regard to the development of the protistan community in temporary aquatic islands (100-1 plastic swimming pools) void of any prior biotic modification. The first experiment ran for 170 d (March–September 1985) and manipulated the access of islands to animals that represent potentially important dispersal pathways for microbes. A second experiment (May–June 1986) investigated in greater detail the early stages of species accrual in the absence of dispersal pathways other than via the atmosphere. Polyurethane foam substrates were used as sampling devices to increase the accuracy and replicability of sampling and provide habitat for colonists. Sampling was determined to be asymptotic. Species accrual was asymptotic in both experiments, although it initially lagged behind that predicted by theory. Autotrophs approached equilibrium faster than heterotrophs, but at a lower species richness. The predicted number of autotroph species at equilibrium was lower in islands that were excluded from contact with birds compared to less exclosed islands. The colonization dynamics of the entire community was not significantly different among islands having different degrees of exclosure. In both experiments, rates of species immigration were nonmonotonic with respect to time and species richness. This relationship appeared to reflect the importance of species interactions during the initial accrual phase before equilibrium. Rates of extinction were positively correlated with both these parameters, although they tended to decrease with time during the equilibrium period in less exclosed islands. Turnover at equilibrium was significant and resulted in directional changes in species composition over time. Assortative processes appeared to be important since later colonists exhibited greater persistence. Colonizing species generally fall into three autecological categories: 1) those that were ubiquitous and had temporally predictable patterns of immigration (successional species); 2) those possessing temporally predictable distributions but not spatially ubiquitous distributions (dispersal limited species); 3) those that showed little temporal or spatial predictability in immigration (transient or allochthonous species). Individual islands exhibited various degrees of fluctuation in species number during the predicted equilibrium which were poorly correlated with exogenous environmental variables and physicochemical habitat parameters. The presence of predacious mosquito larvae(Culex spp.) invariably resulted in a sharp decreases in microbial species richness, while documented contact with rodents was followed by an increase in species number in an island so contaminated. Several aspects of microbial colonization of temporary waters that contradict equilibrium predictions appear to be strongly influenced by microbe-microbe as well as macrobe-microbe interactions.