Common merganser (Mergus merganser) breeding pairs and broods were censused on eight coastal streams on Vancouver Island to evaluate the relative importance of the number of potential breeding pairs, stream size, and the availability of juvenile Pacific salmon in limiting merganser breeding density. The number of potential breeding pairs did not limit nesting density on at least one stream where juvenile salmon populations were enhanced by a hatchery and spawning channel. At distances > 1 km above tidal influence, breeding pairs were evenly dispersed along the streams during the peak egg-laying and incubation period at maximum densities of 0.4–1.4 pairs/km. Maximum breeding pair counts were higher and more variable among streams on the lower kilometer of freshwater (1–9 pairs) and on tidal waters near the stream outlets (4–9 pairs). Stream size accounted for only a part of the variation in breeding pair counts. The estimated number of broods produced on each stream was highly correlated (r = 0.95) with both drainage area and juvenile salmon production, including production from hatcheries. Possible mechanisms relating the dispersion of breeding pairs to the availability of juvenile salmon are discussed. A "food assessment" hypothesis, whereby breeding pairs choose a nesting stream on the basis of prey availability during the nesting season, could not be rejected on the basis of predictions about hatching dates and duckling survival. This hypothesis has serious implications for mortality of wild salmonids in hatchery-enhanced streams.