The minimum requirement of lyophilized plasma samples for a reliable I
nternational Sensitivity Index (ISI) was assessed by calibrations base
d on reducing numbers from a maximum of 60 depleted and 20 plasma samp
les from patients receiving coumarin using a manual technique and low
ISI thromboplastin. The probability of achieving an international norm
alized ratio (INR) result within a clinically important range of 20% d
eviation has been assessed at European Concerted Action on Anticoagula
tion (ECAA) national laboratories in calibrations of ECAA reference th
romboplastin against the certified prothrombin time (PT) with a refere
nce thromboplastin. With conventional orthogonal regression using a ce
rtified PT and linear regression using certified INR, deviations and t
he coefficient of variation of the calibration slope increased with re
duced numbers. The INR deviations became marked when the number of abn
ormal plasma samples was reduced to fewer than 20. Calibrations of 3 a
bnormal plasma samples and 1 lyophilized normal plasma sample gave a h
igh incidence of deviations greater than 10% with an INR of 3.0. The s
tudy demonstrates that with both methods of analysis, an optimum minim
um number of lyophilized plasma samples is needed for a reliable local
ISI.