Researchers have proposed that semantic processing involves both stored semantic knowledge and mechanisms used to access and manipulate this knowledge. Support for this distinction has come from a contrast of the patterns of behavior in comprehension-impaired stroke patients (SA) and patients with semantic dementia (SD). Specifically, SD patients are argued to have a loss of amodal semantic representations whereas SA patients have an impairment of executive control used to access and manipulate these representations (Jefferies & Lambon Ralph, 2006). However, the conclusions of these researchers are based on incomplete and sometimes vague evidence. They have used questionable executive function measures, have failed to fully test semantic dementia patients on measures of executive function, and have provided only weak evidence for a role of executive function in semantic deficits for stroke patients. The current study re-examines the difference between these patient groups and the nature of their deficits using a set of tasks to measure semantic and executive function abilities and the relationship between them. Our results show that the two patient groups are not as distinct as previously claimed. The SD group shows frequency effects only by some measures, while the SA group shows normal frequency effects; both groups show similar item consistency, correlations of semantic task performance regardless of task demands, and some consistent executive deficits. Additionally, we tested stroke patients who showed executive deficits but not semantic deficits, which calls into question the relation of "access" deficits to executive function. We also examined this relationship based on semantic-executive correlations. While our findings do not rule out the possibility of a distinction between knowledge of and access to semantic knowledge, they do call into question some of the main empirical claims used to support this distinction.