Rangeley (1994) has argued that sampling biases, errors in experimental design and low statistical power make the conclusion by Black & Miller (1991), that there was no large impact of an experimental harvest of the seaweed Ascophyllum nodosum on fishes, unmerited and misleading. Our study was conducted mainly at two sites close to Lower Argyle, southern Nova Scotia, Canada where harvesting of Ascophyllum nodosum has been carried out continuously for 30 years along-side intense and productive shellfish and finfish fisheries. In southern Nova Scotia, on-going removal of Ascophyllum nodosum with mechanical harvesters is about 17% (Sharpe & Semple 1991); planned annual removal in New Brunswick where Rangeley has worked will be only about 5% (Anon. 1993). The context of our investigation was a small-scale (400 m2) but extreme, experimental removal (100%) of the seaweed. Our focus was on fishes that moved into the intertidal zone as the tide rose in the evening and early morning and were caught in trammel nets, and on the fishes that were retained in a fixed seine that was set at high tide in early morning and the fishes removed at the next low tide. Thus, subject to the constraint on the size of the fishes imposed by the nature of our fishing gear with its stretch mesh of 25 mm, we were concerned with the identity and abundance of fishes that moved in and out of the intertidal zone and their behavioural response to the 400 m2 patches of habitat that formed our experimental units. The most striking result was the unexpected, low abundance of fishes caught in our gear.