||The USP rabbit bioassay and Limulus amebocyte gel tests were compared for ability to detect purified endotoxin or autoclaved suspensions of Gram-negat1ve and Gram-positive bacteria. The USP rabbit test, the only legally accepted criterion for judging the presence or relative absence of endotoxin from parenteral solutions, could detect as little as 0.01 ug of endotoxin In any volume up to 10 ml/kg as well as 10 /ml Salmonella typhimurium and Escherichia coli, 10/ml Pseudomonas aerugjnosa , or 10/ml Staphylococcus aureus , when the sterile suspensions were injected, 10 ml/kg. The Limulus test could detect as little as 0.00001 ug (0.1 pg) of purified endotoxin/ml or sterile suspensions containing 10 S. typhimurium, E. coli , or Ps. aeruginosa, or 10 S. aureus . Comparative assays of both procedures demonstrated that |he Limulus test is approximately 1000 times more sensitive to purified endotoxin or endotoxin containing bacteria than the legally accepted USP rabbit test. Techicium-albumin preparations, deionized, distilled, and tap water, and catheter washing samples were tested by both USP and Ljmulus methods. In all cases, samples tested by the USP rabbit procedure were negative. All catheter washing samples were negative by Lfmulus test, but tap and distilled water samples were uniformly positive. Deionized water and technicium-albumin samples gave both positive and negative reactions. Since the rabbit and man have been shown to be approximately equal In their response to threshold pyrogenic doses of endotoxin, and since no patient reactions attributable to endotoxin were noted after injection of the technicium-albumin preparations, the Limulus test in unmodified form appears to be too sensitive for judging whether or not solutions Intended for human use are contaminated with endotoxin. Two reproducible techniques for reducing the sensitivity of the lysate have been presented. Simple dilution of active lysate from 1:2 through 1:5 produces a geometric reduction in lysate reactivity. The sensitivity of the lysate can also be reduced by reading the reaction at time intervals prior to the customary 2 hours. Polyacrilamide gel electrophoresis assays of Limulus amebocyte lysate indicate that the lysate is heterogenous, Preparations contained at least 10 distinct protein entities with molecular weights ranging from 22,000 to 185,000. Analyses of several different Ljmulus lysate lots from 3 sources, demonstrated variation in protein content and reactivity to endotoxin unrelated to total protein. Standardization of all lots of Limulus lysate against knov>rn concentration positive and absolute negative controls was found to be of critical importance to insure reproducible results in endotoxin testing.