The stability of tobramycin in pooled human serum when combined with ampicillin, carbenicillin disodium, or penicillin G potassium after storage at 0, 23, or 37 degrees C was evaluated. Samples of pooled human serum containing tobramycin sulfate 8 micrograms/ml alone or combined with ampicillin, carbenicillin disodium, or penicillin G potassium 200 micrograms/ml were prepared and stored at 0, 23, and 37 degrees C. Single samples were removed periodically for 48 hours and frozen until assayed. Tobramycin concentration was measured by a radioenzymatic assay. A tobramycin degradation rate constant was calculated for the tobramycin control and each tobramycin-penicillin combination at each temperature; from this, the time for the tobramycin concentration to decline to 90% of the initial concentration (t90) was estimated. Stability of the penicillins was not assessed. Tobramycin degradation approximated a log-linear process in all samples for the 48-hour period. The tobramycin control sample was more stable than any of the tobramycin-penicillin solutions at each temperature. At 0 degrees C, tobramycin mixed with ampicillin was the least stable of all mixtures; at 23 and 37 degrees C, tobramycin mixed with carbenicillin was the least stable. Storing tobramycin and carbenicillin samples on ice (0 degrees C) prolonged t90 from 10 hours (23 degrees C) and 12 hours (37 degrees C) to 36 hours. The t90 values for tobramycin when mixed with ampicillin were 19, 16.5, and 20 hours at 0, 23, and 37 degrees C, respectively. Mixed with penicillin G, tobramycin t90 values at 0, 23, and 37 degrees C were 48, 44, and 16 hours, respectively. More than a 10% loss of tobramycin potency occurred in some tobramycin-penicillin solutions under the conditions of this study. Because this loss would affect the accuracy of tobramycin pharmacokinetic calculations, the authors suggested guidelines for handling tobramycin serum samples.