Background and aim Acute stroke units in hospitals are known to be more costly than standard care, but proponents claim that the health gains will justify the expense. Yet, despite widespread adoption of stroke units, the evidence on the cost effectiveness of stroke units has been mixed, due in part to differences in the pathway of care across hospitals. The purpose of this study is to compare costs and outcomes for patients admitted to a stroke unit with those admitted to a general ward. Methods Data on 530 stroke sufferers from a large incidence study of stroke (the Auckland Regional Community Stroke Outcome Study) were used. Cost of health services, places of discharge were identified at one-, six- and 12 months poststroke and were linked with long-term cost and survival five-years poststroke. A decision analytical model was developed, including the relationship between waiting time for discharge and probability of admission to stroke unit. Cost effectiveness was determined using a willingness to pay threshold of NZ$20000 (US$15234). Results Regression analysis suggested that there were no significant differences between patients admitted to a stroke unit and a general ward. The incremental cost-utility ratio for the first-year was NZ$42813/quality-adjusted life year (US$32610/quality-adjusted life year), but fell substantially to NZ$6747/quality-adjusted life year (US$5139/quality-adjusted life year) when lifetime costs and outcomes were considered. Probabilistic and one-way sensitivity analysis suggests that the results are robust to areas of uncertainty or delays in the pathway of care. Conclusion Stroke unit care was cost effective in Auckland, New Zealand.