Larger nymphs within aquatic insect taxa have been frequently observed to be transported down-stream in the stream drift only at night. Others have hypothesized this pattern results primarily from large nymphs' behavioural avoidance of entering drift during daylight, when size-selective, visually-feeding fish predators are most active. This hypothesis assumes that animals can actively control their entry into the drift, which may not be the case under all flow conditions. We experimentally induced streamflow increases and decreases in adjacent riffles in a hydrologically-stable stream during the daytime to examine whether changes in diel patterns of drift abundance and size-distribution of mayflies were consistent with the hypothesis of active avoidance of diurnal drift. We assessed the likelihood of active vs. passive mechanisms of diurnal drift entry and transport for four taxa that differ with respect to body size, morpho-behavioural attributes, microhabitat use, and general propensity to drift. In each of three seasons, diurnal and nocturnal drift samples were collected in three riffles over two diel cycles. Background drift patterns were established on the first day (no flow manipulation). Six h before sunset on the second day, flow was experimentally increased in one riffle, decreased in the second, and not altered in the third (control). Between-day differences in diurnal and nocturnal drift rate and size composition were then compared among the treatment and reference riffles. Responses of two taxa were consistent with active control over drift entry, transport, or both. For Baetis spp., drift-prone mayflies typically preyed upon by fish, diurnal drift rates immediately increased following both flow reduction and flow elevation in all seasons, but only small individuals comprised the drift. Drift by large individuals was delayed until nighttime. Epeorus longimanus also exhibited significant increases in drift rates following flow reduction and elevation, but responses of this large-bodied species were restricted to nighttime. Drift responses for these two taxa were largely independent of direction of hydrologic change, thus indicating a strong behavioural control over drift. By contrast, numbers and sizes of drifting Paraleptophlebia heteronea and Ephemerella infrequens depended strongly on direction of flow change. Drift rates for both species generally declined after flow reduction and increased after flow elevation. Moreover, after flow elevation, larger individuals often drifted diurnally, a finding consistent with expectations under a passive hydrodynamic model. These experiments indicate that size-dependent mayfly drift reflects not only presumed risk from visual fish predators, but also functional attributes of species such as morphology, behaviour, and microhabitat affiliation, which influence aspects of drift entry and transport under variable hydrologic conditions.