I examined how constant reduced discharge influenced emigration, abundance, growth, production, and distribution of underyearling rainbow trout Salmo gairdneri in small, seminatural experimental streams in New Zealand. Three stream channels had constant discharges of 0.125 m3/s from mid-November (spring in southern hemisphere) to mid-February. Discharge was then reduced in two channels, but remained unchanged in the third. Reduction of discharge by 32% (to 0.085 m3/s) and by 60% (to 0.050 m3/s) caused no increase in downstream emigration, the emigration rate immediately before reduction being 0.1–0.7 fish/d and that for 5 months after reduction being 0.0–0.3 fish/d. However, the results may have been influenced by unequal population sizes in the channels before the discharges were altered. Presumably, these differences were caused by unequal hatching success of eggs planted in the channel substrates. Population density was greatest (41/100 m2) in the channel selected for lowest discharge and least (19/100 m2) in the channel selected for control discharge. Following discharge alteration, this pattern continued (lowest discharge: 21/100 m2; control: 7/100 m2). Fish size differences between channels at the end of the experiment (7 July) were in direct relation to channel discharges and in inverse relation to population density (largest fish in the control and smallest fish in the low discharge channel). Reductions in instantaneous growth rates were attributable to reduced discharge, but occurred about 2 months after discharge reduction. Reduced discharge depressed production after about 1.5 months and by the end of the experiment there had developed a direct relationship between discharge and production; production was 1.4 g/100 m2at 0.125 m3/s and -3.8 g/100 m2at 0.050 m3/s. There were no significant differences among the channels in the distribution of fish among riffle, pool, and run habitat types. Before discharge reduction, fish in all channels occurred mostly in the riffles and, after reduction, fish in control and treated channels became progressively more evident in the pools. Discharge reduction had no effect on this distribution.