Forest birds are sensitive to habitat change and may be suitable for measuring responses to retention forestry. I present the short-term effects of 6 treatments in a 6 block randomized design experiment on 15 breeding small bird species in mature Douglas-fir (Pseudotsuga menziesii) forests in western Washington and Oregon, 1994–2001. The 13-ha treatment stands contained aggregated (A) green-tree retentions of 100, 75, 40 and 15%, and a dispersed tree distribution pattern (D) of 40 and 15% retention. I compared numbers of entire singing bird territories (abundances) mapped inside sampling plots in post-(2-y x̄) and pre-treatments (1 y). Species richness significantly declined in both 15% treatments. In the 15%D treatment, significantly lower richness and lower species similarity were less than A treatments with ≥40% retention, and significantly lower species diversity was less than the other treatments. The size of decline of abundances of canopy-associated species (summed members) increased with successively lower tree retention; the cavity-nesting species declined with lower snag retention in treatments ≤40% retention; and there was no response for species associated with understory vegetation, but medians of the percentage change in abundance of understory species were negative in cut treatments. There was no detectable difference in treatment effects between tree distribution cut patterns. Greater amounts of tree retention helped maintain composition and abundance better than less retention, and overall the variety of treatments maintained all species. For maintaining richness, similarity, and diversity, the 15%A and 40%D treatments were transitional between the A treatments ≥40% retention, where these community parameters were maintained, and 15%D, which did not maintain natural diversity or species persistence.