OBJECTIVE: To compare the cost and consequences of a policy of continuing to care for patients with a prolonged stay in the ICU with a proposed policy of withdrawing support. DESIGN: Economic evaluation using data derived from a prospective cohort study. SETTING: Adult medical/surgical ICU in a tertiary care hospital. PATIENTS: Consecutive patients admitted to the ICU. INTERVENTION: None. MAIN OUTCOME MEASURES: We performed a cost-accounting analysis on each patient in the ICU and followed up patients until 12 months after admission to ICU and assessed components of quality of life in survivors. RESULTS: During the study period, 690 patients were admitted to the ICU. Only 61 (9%) patients remained in the ICU for > 14 days. For this group, the mean length of stay in the ICU was 24.5 (+/-11.7) days and duration in hospital was 57.9 (+/-56.9) days. At 12 months, 27 (44%) were alive. Overall, the mean quality of life score at 12 months did not differ between patients with a short or prolonged stay in the ICU. The average ICU cost per day per patient was $1,565 (Canadian) resulting in a total cost for the whole cohort of Can $1,917,382. Over the same time period, 58 patients had life support withdrawn. On average, patients survived another day in the ICU, 2 more days in hospital, and all patients ultimately died. When treatment was discontinued, the costs of treating this cohort was Can $156,465. The incremental cost-effectiveness ratio is Can $65,219 per life saved or Can $4,350 per life-year saved. CONCLUSIONS: A considerable proportion of patients with a prolonged length of stay in the ICU survive their critical illness. Furthermore, their long-term quality of life seems reasonable. Our data suggest that continuing treatment for patients with a prolonged ICU stay may represent an efficient use of hospital resources and should be considered in the context of local budgets.