Prothrombin time was measured in three different plasma samples by 2580 laboratories in the 1977 CAP Proficiency Testing Program. Analysis of variance was used to show that instrument, as well as thromboplastin, has a significant effect upon observed prothrombin time. In addition, the instrument and thromboplastin effects were estimated, and all were shown to be linearly related to the prothrombin time of the plasma sample. This linear relationship was used to develop a formula for adjusting/correcting an observed prothrombin time for both the thromboplastin and instrument effect. This adjustment/correction method seems promising on the basis of its use in four different sets of data.