The success in achieving therapeutic international normalized ratio (INR) targets in the control of warfarin using a whole-blood point-of-care testing (POCT) monitor (CoaguChek) in a community clinic was compared with hospital laboratory coagulometer prothrombin time (PT) testing in a randomized crossover study. Forty-six patients were randomized into two groups. At each visit, capillary blood was taken for the POCT monitor and venous blood for the laboratory coagulometer. In Group 1, for 6 months, dosage was based on the CoaguChek and for the second 6 months on the coagulometer. In the second group, the order was reversed. Dosages were determined using the dawn ac computer programme. Success was assessed by the percentage of time patients were maintained within the INR targets. Agreement between laboratory and monitor INR, and patient satisfaction were also assessed. Results with the POCT monitor compared well with the hospital coagulometer. Time in INR target range between the groups was similar, with 60.9% on the POCT monitor and 59.3% with the laboratory coagulometer in Group 1 and in Group 2, respectively, 64.3% and 63.4% with no significant difference in mean INR. An INR above 4.0 gave some discrepant results. International Sensitivity Index calibrations of the two test systems indicated that the INRs were dependable. Patient questionnaires showed greater satisfaction with community POCT monitoring.