A severe wildfire burned 454 hectares of a second-growth Douglas-fir forest in 1977 on a north-facing slope in Pattee Canyon, near Missoula, Montana. The slope was aerially seeded with a grass mixture, from which Dactylis glomerata established best. Community structure, conifer regeneration, and the impact of the seeded grass on the plant community were evaluated with two data sets. One set tracked postfire vegetation development from 1979 to 1987 on permanent transects established on upland sites in areas of varying fire severity. The second data set was collected in 1989 on upland sites nearest to the bum edge where conifer regeneration was expected to be greatest. Most stands converged to a similar ordination space by 1987, showing that several key species which established in the initial postfire year determined community structure. These species were largely on-site survivors (including Calamagrostis rubescens, Physocarpus malvaceus, and Spiraea betulifolia) and the seeded grass, D. glomerata. The spread of this species by 1987 to stands that apparently escaped seeding in 1977 suggested that this species may persist at this site. Patterns of species abundance and distribution in 1989 were primarily controlled by factors summarized by a topographic-moisture index and by pre-burn disturbance history. Three general site types were described by an ordination. Areas that appeared as open woodlands in 1937 occurred on ridges and had more xerophytic vegetation in 1989, including native grasses. Conifer regeneration in this region was limited primarily to Pseudotsuga menziesii (Douglas-fir). Areas that appeared as an even, young forest in 1937 were on open slopes and were associated with Pinus contorta (lodgepole pine), Larix occidentalis (western larch), Vaccinum globulare, and Amelanchier alnifolia. Areas that appeared as an older, uneven-aged forest in 1937 were on open slopes near the upper burn edge and were characterized by Douglas-fir and Spiraea betulifolia in 1989. D. glomerata was more successful on the drier ridges and was negatively associated with Calamagrostis rubescens. Competition with C. rubescens, rather than differences in environmental tolerances, most likely restricted D. glomerata to the more xeric sites. An adverse effect of D. glomerata on conifer regeneration was most likely for western larch on xeric sites. Site factors and historical factors were most important in determining patterns of conifer regeneration. Regeneration was moderate on mesic slopes and sparse on xeric ridges. The extreme density of lodgepole pine regeneration (13,000 stems/hectare) in one mesic area reflected the importance of serotiny for post-fire regeneration of this species. Douglas-fir regeneration ranged from 370 stems/ha in a mesic area where no mature survivors were noted to 4045 stems/ha on a mesic slope near survivors. On mesic slopes near survivors, western larch regeneration was 857 stems/ha and was minimal elsewhere. A regression model confirmed the importance of site factors, site history, and availability of seed source for Douglas-fir and lodgepole pine. The model for western larch was only able to explain 14% of regeneration pattern, suggesting that microsite variation as well as other variables would be needed to predict regeneration for this species. / Graduation date: 1992