Stem-mapped plots in old-growth forests of western hemlock ( Tsuga heterophylla) and western redcedar ( Thuja plicata) in northern Idaho, USA were analyzed using Ripley's K( d) function, nearest-neighbor function, and influence zone analyses. A conceptual model of old-growth forest development was formulated from the spatial pattern analyses, to guide the development of a mathematical model. In the conceptual model, cohorts of seedlings begin life established in clusters associated with canopy gaps created by the deaths of overstory trees. Then, as the trees within clusters increase in size, they begin to compete with their immediate neighbors. Density-dependent mortality thins the clusters and increases the distance between neighboring trees. Over time, this self-thinning behavior tends to drive stand spatial patterns from aggregation towards regular spacing as trees get larger or increase in competitive status. Preliminary results from a dynamic point process model are presented. The approach simulates the regeneration of seedlings in gaps and the dynamic spatial patterns resulting from competitive interactions between neighboring trees as a sequence of point processes. Main features of the model are stochastic assignment of gapmaker trees, a Poisson cluster process for regeneration establishment, and a progressive simple inhibition process for competition between neighboring trees. The model produces spatial patterns for regeneration and adult trees consistent with the conceptual model and with patterns observed in the field data. Refinements designed to improve model realism are discussed.