The longleaf pine (Pinus palustris Mill.) forests are important ecosystems in the southeastern USA because of their ecological and economic value. Since European settlement, longleaf pine ecosystems have dramatically declined in extent, to the degree that they are now listed as endangered ecosystems. Its sporadic seed production, which limits the frequency of natural regeneration, is identified as a significant factor in this decline. Previous studies did not characterize the complexity in cone production. Here a method of multiscale entropy is used to analyze long-term data for cone production in longleaf pine forests at six sites across its native range. Our results indicate that there exists a regime shift for cone production at each site. The corresponding time scales of the regime shift are generally 1–9, 10–12, 13–16, and 17–24 yr. Overall patterns for the complexity of cone production with the change of time scale are similar among sites, with exception of the Red Hills (FL). There are high correlations between entropy of cone production and entropy of annual mean air temperature and annual total precipitation at all sites. These results provide new insight into the complexity of cone production of longleaf pine forests with significant management implications.