OBJECTIVE: To review approaches to assessing consent capacity in patients with neurocognitive or neuropsychiatric illness; to summarize the rationale behind our structured interview for consent capacity; and to outline questions for future research. METHOD: After reviewing legal and clinical literature, and empirically comparing three leading consent capacity instruments, we developed the Assessment of Capacity to Consent to Treatment (ACCT) interview and administered it to adults with dementia (n=20), schizophrenia (n=20), and controls (n=19). Capacity ratings by primary care clinicians and experts blind to the patients' status were obtained for a subsample. RESULTS: Interscorer reliability was r=.90; internal consistency reliability was .α=96. ACCT scores agreed 82% of the time (kappa = .44; p<.01) with primary care clinician ratings of capacity and 75% of the time (kappa = .50; p<.05) with expert ratings of capacity. Patients with dementia and schizophrenia could express treatment choices but performed worse than controls on measures of understanding, appreciation as problems with foresight, rational reasoning, and values-based reasoning. Only patients with schizophrenia performed worse on a measure of appreciation as problems with distrust. CONCLUSION: The method of assessing consent capacity described here has adequate reliability and validity, and may provide a useful starting point for clinicians and researchers. Many questions remain about the nature of consent capacity, its component constructs, and the meaning of instrument versus clinician ratings of capacity. Future adaptations, particularly in the assessment of appreciation and reasoning, and additional studies in more diverse samples, are needed.