OBJECTIVE To compare a new low-dose warfarin induction regimen with the Fennerty regimen in elderly inpatients. METHODS Age-stratified, randomized prospective study. METHODS 120 age-stratified elderly inpatients. METHODS Each patient was randomized to either the new induction regimen or to a modified Fennerty regimen. METHODS Days to therapeutic International Normalized Ratio (INR >2); days in the therapeutic range (INR 2-3) during induction; number of patients with INR >4.5; ability of day 4 INR to predict day 8 warfarin dose. RESULTS The mean time to therapeutic INR was longer for the new induction regimen than modified Fennerty regimen in patients aged 65-75 years [4.6 (mean) +/- 1.6 (SD) days vs 3.8 +/- 0.8 days; P = 0.03] and in patients aged >75 years (4.5 +/- 1.4 days vs 3.5 +/- 0.7 days; P = 0.003). Patients spent more time in the therapeutic INR range with the new induction regimen [3.0 +/- 1.3 days vs 2.7 +/- 1.3 days (P = 0.03) for those aged 65-75 years and 2.9 +/- 1.1 days vs 2.4 +/- 1.3 days (P = 0.04 for those aged >75 years]. Fewer patients using the new regimen had INRs >4.5 in the first 8 days [1 (3%) vs 6 (20%) for 65-75 years (P < 0.05) and 1 (3%) vs 11 (37%) for >75 years (P < 0.01)]. The ability to predict the maintenance dose to within 1 mg was 55% for both regimens. CONCLUSIONS The low-dose regimen has important clinical advantages over the Fennerty regimen for anticoagulating elderly inpatients.