A test for determining the sterility of a total nutrient admixture (TNA) containing equal quantities of 10% fat emulsion (Liposyn II), 8.5% amino acids injection, and 50% dextrose injection using the USP membrane filtration procedure was developed and evaluated. Membrane filter selection was determined by analysis of flow rates, membrane fluid compatibility, bubble point stability, and rinse fluid requirements. Microbial challenges employing five organisms (Bacillus subtilis, Escherichia coli, Candida albicans, Staphylococcus aureus, and Pseudomonas aeruginosa) and both soybean casein digest and fluid thioglycollate media were used to confirm the ability of the test to detect low-level microbial contamination. A polyvinylidene fluoride membrane was determined to be the most appropriate of the membrane types studied because of its superior flow rate and membrane-fluid compatibility. Bubble point testing revealed no detrimental effects on the membrane. The potential problem of haziness caused by retention of the TNA by the membrane with subsequent release in the culture media (which could result in false-positive growth determinations) was diminished by using a sterile 0.1% peptone solution rinse and careful observation techniques. Performance of the sterility test by six hospital pharmacists required an average of 14.2 minutes. Sterility testing of alternate TNAs compounded with Intralipid and Nutralipid was not feasible because of prolonged filtration times. The basic USP membrane filtration procedure for large-volume injections can be used by hospital pharmacists for testing the sterility of TNAs. When fat emulsions are used in compounding, sterility-testing procedures specific to the emulsion product used should be developed and evaluated.