International normalized ratio (INR) discrepancies were noted between clinical laboratories using various prothrombin time (PT) systems. We studied the influence of different commercial blood collection tubes and different PT systems on INR measurements. INRs of fresh patient samples were determined by 3 laboratories, each using different PT systems. In the first part of the study, samples were drawn with Vacutainer tubes and in the second part with Monovette tubes. In the first part of the study, the maximum bias for all patients amounted to 0.46 INR (14%), and in the second part, to 0.14 INR (4.9%). The maximum bias for all patients could be reduced further by local system calibration using frozen pooled plasma specimens. The sodium citrate solutions in the blood collection tubes were contaminated with magnesium ions (approximately 2.7 mmol/L and 0.3 mmol/L in the Vacutainer and Monovette, respectively). INR discrepancies could be explained largely by this influence of blood collection tubes. The maximum allowable magnesium contamination in sodium citrate anticoagulant solutions should be less than 1 mmol/L.