This article describes a new computational model of aphasic sentence comprehension. The model is based on the premise that all aphasics, however different, share a common deficit which determines a considerable amount of the individual variation observed in their sentence comprehension performance. This common deficit is construed as a pathological reduction in the activation resources of a working memory system that subserves sentence comprehension (Miyake, Carpenter, & Just, 1994). To test the theoretical feasibility of the resource reduction hypothesis, a new computer model of aphasic sentence comprehension was developed and tested. We describe the model as well as some initial simulation results, indicating that the model can account for some of the sentence complexity and severity effects that have been reported in the aphasia literature.