The distribution of egg masses throughout the environment can strongly influence the population dynamics of aquatic insects. Many species lay eggs in aggregations and most eggs will subsequently hatch from only a few locations-knowing how and why these aggregations arise is therefore needed to understand the population dynamics of these species. The hydrobiosid caddisfly Ulmerochorema rubiconum lays eggs in large aggregations on the undersides of emergent rocks in streams. Our aim was to test whether females oviposit in response to conspecific egg masses or the environmental characteristics of rocks by manipulating the number and age of egg masses initially present on rocks and monitoring the accumulation of new masses. Our first experiment used rocks that had recently been used for oviposition and initial egg masses encompassed a range of ages; our second used rocks with no history of oviposition and initial masses of uniform age. Females did not respond to conspecific eggs of any age, as removing initial egg masses had no impact on the number of new masses laid in either experiment. There was a strong positive relationship between the number of initial and new masses, regardless of whether initial egg masses were removed, at the scale of whole rocks and for rock microhabitats. The results suggest that females select oviposition sites based primarily on physical characteristics of emergent rocks. We demonstrate for the first time that the spatial arrangement of egg masses may be consistent and predictable through time, with females showing a high degree of fidelity for particular rocks and microhabitats.