This study investigated comprehension of wh-questions in two Broca's aphasics. Patients were presented for comprehension with two types of wh-questions: questions headed by which and questions headed by who. These two types were chosen because according to recent syntactic analyses they give rise to different types of syntactic "chains." These questions were presented in both subject gap versions (e.g., which cat chased the dog?) and object gap versions (e.g., which cat did the dog chase?). Comprehension of which questions was asymmetric, with subject gap versions comprehended significantly better than object gap versions, the latter yielding chance-level performance. This finding is consistent with previous reports of subject-object asymmetries in comprehension of relative clauses and clefts, as well as active-passive comprehension asymmetries. In contrast, comprehension of who questions was symmetrical over subject gap and object gap versions: Both patients performed equally well (significantly better than chance) on subject gap and object gap who questions. These findings are inconsistent with current formulations of "chain" or "trace"-based theories of agrammatic comprehension which assume a deficit that affects both types of syntactic chains. We suggest that linguistic descriptions of agrammatic comprehension should be limited to deficits involving only one type of chain. We also suggest that there are processing differences underlying the syntactic distinctions between which-type and who-type questions and that this may account for different patterns of comprehension on these and other constructions.