The importance of density-dependent mortality in altering forest structure was assessed by examining the effects of mortality and recruitment on spatial arrangement in thirteen plots in an old-growth conifer forest. The effects of each process over a twelve to twenty-three year period were analyzed both individually and in combination at two scales. Changes in spatial pattern attributable to density-independent mortality were almost as common as changes attributable to density-dependent mortality, indicating that mortality processes were not dominated by density-dependent effects in the forest as a whole. In addition, density-independent mortality was not always random in its effect, belying a simple dichotomy between density-dependent and random mortality. Furthermore, changes due to recruitment frequently counteracted changes due to density-dependent mortality. An analysis of spatial pattern for large (diameter at breast height≥60cm) and small (diameter at breast height<20cm) trees was also performed. Trends attributable to density-dependent mortality were even less frequent within size class than for the population as a whole. This was particularly true for the large trees, where changes due to mortality at the smaller spatial scale generally showed no trend and at the larger spatial scale frequently showed trends toward increased clumping. Finally, a spatial analysis for broad types of mortality factors was performed. The category of mortality factor most likely to contain density-dependent effects  showed, as expected, a strong tendency toward clumping, while the category most likely to contain density-independent effects showed either clumped or random patterns. In summary, at least for these old-growth stands, one cannot rely on a simple competition-based theory for how forest structure will change due to the action demographic processes.