An observed point pattern consisting of the locations of trees within a forest stand should be representative of the ecological processes acting during stand development. To analyze point patterns, researchers typically rely on testing uninformative null models such as complete spatial randomness (CSR). Unfortunately, the inference available from tests using such models is limited, and instead, more informative spatial point process models should be used. The use of such models is complicated, and a literature review shows that in general they have been ineffective at providing answers about ecological processes, largely as a result of methodological issues. My thesis is that spatial point process models can be useful to elucidate ecological knowledge, as long as the research (i) uses an analytical method that incorporates exploratory data analysis (EDA), model specification including alternative model forms, parameter estimation, and model assessment potentially including both multiple distance statistics, and power sample size analysis, (ii) occurs within a framework that ensures the proposed research question is answerable based on limitations of both the data and analytical methods, and (iii) focuses on learning about the type, strength and scale of the observed spatial structure, along with the possible existence of an underlying trend surface. I provide two examples demonstrating how fitted models, including both the parameter values and the model structure, can be related to ecological process. In the first example I consider an inhibition model applied to mortality data and in the second, a cluster model applied to seedling establishment data. Finally, I demonstrate that how CSR envelope is inconsistent at detecting the so-called spatial scale of a process, and also that the K statistic alone is insufficient for assessment of spatial point process models.