Fire-scarred trees provide an important source of data and direct evidence for understanding past fires and vegetation dynamics. Although dendrochronological fire scar records provide fine-scale temporal and spatial resolution, limitations exist in the development, analysis, and interpretation of these fire history datasets. In order to assess these limitations, we applied a systematic gridded sampling scheme to a 4-hectare study area located in the Ridge and Valley Province of northeastern Pennsylvania. We mapped and described a total of 155 fire-scarred trees within a 50-meter resolution grid, noting the species, health status, and basal scar orientation for each tree. Additionally, we cut a full or partial cross section from 58 fire-scarred pitch pine (Pinus rigida) for the purpose of assessing multiple "filtering" techniques and for the development of a fire interval simulation model. The simulation model randomly selected trees from each grid cell to estimate fire return intervals at multiple spatial scales. Our results indicate that fire return intervals are dependent upon the definition of "recorder" years and influenced by various filtering methods, including minimum number and minimum percentage of trees scarred. For example, the number and length of fire intervals was greater when years prior to the initial scar were considered "non-recording" vs. "recording" and when a percentage scarred filter was applied. The simulation model provides an additional range of fire interval estimations that can be used by land managers to guide forest restoration and fire management objectives.