Nine partial thromboplastin (cephalin) reagents have been compared in a parallel investigation of groups of patients on `long-term' anticoagulants, a group with moderate haemophilia, and patients on heparin infusion.Results with the seven commercial reagents and a human cephalin extract have been correlated with those of a specially prepared and standardized reference preparation of human brain origin. The comparison was similar in principle to that of the prothrombin time thromboplastin standardization using the British Comparative Thromboplastin (BCT).Results, which for comparative purposes were expressed as ratio of patients' cephalin times to control cephalin times, varied greatly in all three groups. In the oral anticoagulant group some of the commercial reagents were particularly insensitive to the `intrinsic' clotting defect. The correlation between the `standardized preparation' and the other reagents was not good and the use of a reference cephalin material for quality control of cephalin time tests does not appear promising.In moderate haemophilia the commercial reagents were either relatively poor at picking out the clotting defect compared with the `standardized preparation' or gave such a bad endpoint that the results were not dependable. The poor endpoint also limited the dependability of the results of all but the `standardized preparation' and two of the commercial reagents in controlling heparin administration.In view of these standardization difficulties, which cannot apparently be resolved by the use of reference material, there is need for bulk, routine supplies of a sensitive, standardized cephalin reagent giving good reproducible endpoints. The method for the provision of such material in a recently introduced national supply scheme is described.