At the present time there exists no satisfactory procedure suited to routine application for the quantitative determination of penicillin in blood. The numerous methods which have been described for the assay of penicillin in body fluids are generally deficient in two respects: (1) visual estimation of inhibition is relied upon, and whether tube or serial dilutions are used this procedure results in considerable error (25 to 100 per cent); and (2) since the unknowns are compared with a standard which contains no serum, the questionable assumption that the two series are directly comparable is necessary. For most clinical procedures, these features are not serious objections. In certain instances, however, a quantitative procedure would be very useful. A case in point is the study of the relative values of various modes of penicillin administration. The usual plate or cup methods of penicillin assays are generally inapplicable to this problem. The amounts of penicillin required to produce inhibition in the former, or measurable zones in the latter, are relatively high, considerably in excess of those known to be therapeutically effective. Recently, however, Cooke (1945) has described a modification of the plate method suitable for application to blood. This method is, however, only semiquantitative. A number of broth methods have been proposed (Rammelkamp, 1942; Rosenblatt et al., 1944; Kirby and Rantz, 1944; Fleming, 1945). In all of these methods, the concentration of penicillin is determined by estimation of the amount of fluid required to inhibit either the growth or the hemolytic activity of a hemolytic streptococcus. Kirby and Rantz (1944) have suggested a modification of the Foster (1942) turbidimetric procedure for the determination of penicillin in serum. Nonspecific growth stimulation is avoided by the use of complementary volumes of the patient's serum, obtained prior to penicillin administration. The latter requirement seriously limits the application of the procedure described. In attempting to devise a satisfactory turbidimetric assay for penicillin in blood, a considerable body of information relating to the effect of serum, plasma, and their globulin-free filtrates upon the growth in broth culture of Staphylococcus aureus has been obtained. This paper contains a summary of these data, together with a consideration of their relationship to the problem of such an assay. EXPERIMENTAL METHODS AND RESULTS In some of the earlier experiments, 1.0-ml volumes of the solutions to be tested, suitably diluted when necessary, were mixed with 5.0 ml of broth. Later it was 279