Habitat selection is often viewed as a major process determining the distribution and abundance of animals. If animals have limited control over their dispersal, however, they may sometimes be unable to colonize preferred habitats. In such cases, observed spatial variations in abundance may reflect the vagaries of colonization rather than true habitat preferences. Our study examined whether local variations in the density of stream insects are related to habitat quality or colonization history. We focused on larval black flies (Simulium vittatum), a representative group of suspension feeders that exhibit strong preferences for habitats characterized by high near-bed velocities. Despite these preferences, variations in the density of S. vittatum larvae across benthic substrates within the riffles of a Pennsylvania stream were often weakly related to velocity. Moreover, densities on substrates of similar velocity varied by more than six-fold. We conducted a field experiment to test whether these density differences reflect habitat preferences and employed a short-term behavioral assay based on larval emigration to quantify preference. We used a block design to compare emigration rates from paired substrates that had been colonized by markedly different larval densities despite their similar velocities. After removing these larvae, we added larvae to both substrate types in equal numbers, following which we measured larval emigration. There was no significant difference in emigration between these paired substrates even though statistical power was high, which indicates that the large density differences present at the end of the colonization period on substrates of similar velocity were unrelated to habitat preferences. There were, however, significantly more departures from blocks of substrates in lower velocities than from those in higher velocities, demonstrating that the duration of our experiment was sufficient for black fly larvae to express their habitat preferences. These results are consistent with the hypothesis that dispersal constraints associated with the settlement of drifting larvae limit their ability to reach preferred habitats. In fact, related studies predict that larval settlement rates are lowest in the high velocity habitats preferred by larvae. We conclude that such dispersal constraints may be important for a wide variety of freshwater, marine, and terrestrial organisms that are transported by moving fluids.