This investigation was designed to compare the enzyme-modified immunoassay (Syva--EMIT) with a radioimmunoassay (New England Nuclear--RIA) and the radiometric assay (Johnston--BACTEC) to determine the optimal assay for use in our aminoglycoside dosing service. The serum concentration determinations obtained via the three assay methods were analyzed by linear regression analysis. Significant positive correlations were noted between the three assay techniques (p less than 0.005) during both sample collection phases. The coefficients of determination for EMIT vs BACTEC and RIA vs BACTEC were 0.73 and 0.83 during phase 1, respectively, and 0.65 and 0.68 during phase 2, respectively. The slope of the regression lines also varied markedly during the two phases; 0.49 and 0.42 for EMIT and for RIA vs BACTEC, respectively, during phase 1 compound with 1.12 and 0.77, respectively, during phase 2. The differences noted in these relationships during phase 1 and 2 may be related to the alteration of the pH of the control sera utilized in the BACTEC assay. In contrast, RIA vs EMIT regression analysis indicated that existence of a highly significant relationship (p less than 0.0005 and r2 = 0.90). The EMIT technique was the easiest and most accurate for determination of serum gentamicin concentrations, whereas the BACTEC method was judged unacceptable for clinical use.