A more efficient design, based on a graded response technique, for the bioassay of pituitary extract (posterior lobe) is presented. Eight longitudinal strips of uterine muscle from one virgin guinea pig are employed. Two dosage levels are used, differing by a fixed percentage, for both the standard and the unknown preparations. Doses are added to the bath in a random order. Factorial analysis of the responses gives the most probable potency of the unknown preparation, and the standard error of this observed potency is found by means of the analysis of variance. The method has the advantages of speed and accuracy. As many as four unknown extracts have been assayed in one day. For 23 unknown solutions the mean actual error was 7.7%, and the mean standard error was 3.8%.