Many primary care laboratories use point-of-care (POC) instruments to monitor patients on anticoagulant treatment. The internal analytical quality control of these instruments is often assessed by analysing lyophilised control materials and/or by sending patient samples to a local hospital laboratory for comparison (split sample). The aim of this study was to evaluate the utility of these two models of prothrombin time quality control. The models were evaluated by power functions created by computer simulations based on empirical data from 18 primary care laboratories using the POC instruments Thrombotrack, Coagu-Chek S, or Hemochron Jr. Signature. The control rules 1(2S), 1(3S), exponential weighted moving average, and the deviation limits of +/- 10% and +/- 20% were evaluated by their probability of error detection and false rejections. The total within-lab coefficient of variation was 3.8% and 6.9% for Thrombotrack, 8.9% and 10.5% for CoaguChek S, and 9.4% and 14.8% for Hemochron Jr. Signature for the control sample measurements and the split sample measurements, respectively. The probability of error detection was higher using a lyophilised control material than a patient split sample for all three instruments, whereas the probability of false rejection was similar. A higher probability of error detection occurred when lyophilised control material was used compared with the patient split samples; therefore, lyophilised control material should be used for internal analytical quality control of prothrombin time in primary health care.