Purpose: The goal of this study was to identify some potential key cognitive and communicative processes underlying narrative discourse ability following traumatic brain injury (TBI). Specifically, this study (a) investigated the contribution of working memory (WM) and inferencing to narrative discourse comprehension and production; (b) tested key assumptions posited by the Structure Building Framework (SBF), a discourse model; and (c) evaluated the potential for inferencing to contribute to discourse ability beyond a shared variance with WM. Method: Twenty-one individuals with TBI completed six tasks yielding seven measures: verbal and nonverbal WM updating (WMU-V and WMU-NV, respectively), predictive inferencing, the Discourse Comprehension Test (DCT), a picture story comprehension (PSC) task, and story retelling (story grammar and story completeness). Regression analyses were performed using WM and inferencing as predictors for narrative performance. Results: WM measures were significant predictors of DCT performance and approached significance as predictors of PSC. Inferencing approached significance as a unique predictor for the DCT and story completeness. WMU-V and WMU-NV were highly collinear, and neither WM measure predicted discourse outcomes over and above the other's contribution. Conclusions: WM was more strongly associated with comprehension processes, whereas inferencing may be associated with both comprehension and production outcomes. Findings were interpreted as supporting SBF assumptions of domain generality of cognitive processes and mechanisms involved in discourse while also challenging assumptions that the same cognitive substrates are marshaled for comprehension and production processes.