OBJECTIVE: To estimate the cost effectiveness of screening for LVSD in patients with cerebro-vascular accident (CVA), peripheral vascular disease (PVD) or transient ischaemic attack (TIA) versus age matched controls. METHODS: Results of electrocardiography (ECG) or measurement of brain natriuretic peptide (BNP) were compared with echocardiography (LVSD = ejection fraction <40). Cost effectiveness of ACE inhibitor (ACE-I) treatment of LVSD was estimated from 12 published studies. RESULTS: We investigated 257 cases with CVA, PVD or TIA and 199 controls. The overall prevalence of LVSD was 28% in the cases and 6% in the controls. Past history of MI was associated with a high risk of LVSD in both cases (48%) and controls (33%) (Relative Risk 1.53, 95% CI 0.87-2.71). However, 21% of cases with no past history of MI had LVSD in comparison with 1% of the controls (Relative Risk 17.51, 95% CI 4.29-71.40). Measurement of BNP cost more than ECG screening, was less sensitive and specific and consequently not as cost effective. Echoing only those patients with history of MI was the most cost effective strategy in cases and controls (cost per LVSD detected $786 and $1,203). The cost per additional LVSD detected by ECG screening of patients with no MI was $1,392 in cases vs $22,304 in controls. The cost per additional LVSD detected by echoing all subjects was $3,349 for cases vs $40,488 for controls. Screening and treating LVSD with ACE-I was cost effective (