Sampling efficiency (p), estimated from the Zippin depletion model and the percentage of fish obtained during first runs (single-pass electrofishing), which is also named relative abundance (C1), calculated as the ratio of C1 to estimated density for a site (Ce), were analysed in five contiguous sites in a small lowland stream in Poland over 23 years. The sampled reach was unregulated during the first decade of the study period, but regulated by weirs and sluices in the latter 13 years. The sites were populated with 17 fish species during this period. Sampling efficiency p calculated for all species did not differ significantly between the five sites in any given year, between benthic and non-benthic species, or between the natural and regulated periods. Differences in p were only observed between some species. C1 differed significantly between large and small species and, except for roach, was not dependent on the number of runs. Results indicate a high stability of both sampling indices in this small stream, despite changes in channel morphology after regulation: information which could be useful for future sampling research in streams.