1. Completion of a large interbasin water transfer system in northern Texas (U.S.A.) provided the opportunity to test the effects of pre-planned, experimental increases (??? ??30) in flow on the fish fauna of a small, low-gradient, natural stream that was included as part of the conveyance system. Water from Lake Texoma (Red River basin) was pumped via a 16-km pipeline to the headwaters of Sister Grove Creek (Trinity River basin), which then carried the donor water 50 km downstream to Lake Lavon. 2. Baseline (pre-transfer) data on the composition of fish assemblages at seven stations on the creek or at its confluence with the receiving reservoir were collected monthly for 3 years, and similar data were collected for 2 years during and after trial flows of Lake Texoma water to Sister Grove Creek. We also documented fish abundance at five creek stations immediately before and after three trial flow periods of 10-14 days each in summer and autumn. 3. Multivariate analysis of all routine monthly samples over the 5-year pre- and posttransfer period showed moderate changes in the fish fauna of the creek after initiation of the trial flows. Samples taken within a week before and after the artificial high flows showed little overall change in abundance of individual fish species, but at some stations the quantitative or qualitative change in composition of the local assemblage was substantial. 4. The trial flows lasted 2 weeks or less. Long-term effects of water transfer on the fish fauna of Sister Grove Creek can only be determined after the conveyance system goes into normal operation, with periods of artificial flow of longer duration.