Objective: To determine the impact of a protocol on hospitalization costs for patients admitted with stroke. Design and setting: Nonrandomized control trial in an urban community hospital with 376 beds. Patients: All patients admitted with a diagnosis-related group code of 014 (cerebrovascular disease) were included (N = 390). Patients with subdural hematoma (N = 2) or subarachnoid hemorrhage (N = 2) were excluded. Intervention: A protocol for treatment of acute stroke was developed that included a critical path for nursing care, an algorithm for emergency department care, and suggested admission orders for physicians. Main outcome measures: The hospital information system computer database was searched for hospitalization charges, length of stay, tests performed, and treatments provided. Results: Patients treated with the protocol had lower charges compared with historical (p = 0.026) and concurrent (p = 0.02) control groups. Lower charges were accounted for by a decreased length of stay in the protocol group compared with historical (p = 0.001) and concurrent (p = 0.13) controls. Tests and treatments provided were similar except that carotid Doppler studies and deep venous thrombosis prophylaxis were more frequently done in those treated with the protocol (p = 0.001 for carotid Doppler and p = 0.026 for deep venous thrombosis prophylaxis). There were no differences in outcome measures such as death or discharge disposition. Medical complications were similar in all groups. Conclusions: There were significant savings in hospitalization cost for patients with acute stroke after introduction of a treatment protocol. These savings were almost entirely related to decreased length of stay. The protocol led to modest differences in tests ordered and treatments provided.