Membrane filtration, small-aliquot inoculation, and double-strength broth methods of sterility testing were evaluated for detection of small numbers of bacteria in 5% dextrose injection (D5W). Each of 240 bags of D5W 50 ml were inoculated with approximately 10 2 Staphylococcus epidermidis and subjected to one of the three test methods at 1, 3, 6, 9, 12, 18, 24, or 48 hours after inoculation. After incubating at 25 degrees C for seven days, the test units were examined for turbidity, indicating growth of bacterial contaminants. Double-strength broth was shown to be more reliable than the other two test methods, detecting the bacterial contaminants in 30 of 30 samples through six hours. Successful recovery of low-level Staph. epidermidis in D5W decreased significantly after a nine-hour delay in processing. Membrane filtration and aliquot-sampling methods were comparable, each detecting contamination in 3-4 of 10 bags at one hour after inoculation. The number of false negatives increased with time, with no contaminants detected in any of the bags tested with these two methods nine hours after inoculation. It is concluded that the testing method selected to monitor for sterility and the amount of time elapsed before processing the sample are critical to the accuracy of results.