Within communities, organisms potentially self‐organize through endogenous processes that create nonrandom spatial structure as they interact with one another or modify the abiotic environment. In contrast, exogenous processes such as environmental heterogeneity or variable immigration are thought to be dominant processes controlling these spatial patterns. Although both endogenous and exogenous processes likely occur, their relative importance is still largely unknown because of limited analytical tools and the lack of experimental evidence, particularly those that address exogenous sources of environmental heterogeneity. Here, we used a soil heterogeneity experiment to examine the relative effect of endogenous and exogenous processes on plant spatial structure after five years of community assembly. Soil heterogeneity was manipulated by splitting the vertical soil profile into three soil‐types that were randomly assigned to 40 × 40 cm patches within 2.4 × 2.4 m plots. Homogeneous plots were created by mixing all soils before filling each patch. Thirty‐four grassland species were then sown into all plots and allowed to grow for five years after which the location of all plants was mapped using a 5 × 5 cm grid. Results from point‐pattern spatial analysis indicated that, even in the absence of soil heterogeneity and with initial seed arrival, spatial structure was primarily generated by endogenous processes. Although soil heterogeneity increased species aggregation at certain scales, most of the spatial structure was created by endogenous processes. These results suggest that endogenous processes may be more important than expected for generating spatial structure and can develop much faster than anticipated.