The diagnosis of male hypogonadism requires the demonstration of a low serum testosterone (T) level. We examined serum T levels in pedigreed samples taken from 62 eugonadal and 60 hypogonadal males by four commonly used automated immunoassay instruments (Roche Elecsys, Bayer Centaur, Ortho Vitros ECi and DPC Immulite 2000) and two manual immunoassay methods (DPC-RIA, a coated tube commercial kit, and HUMC-RIA, a research laboratory assay) and compared results with measurements performed by liquid chromatography-tandem mass spectrometry (LC-MSMS). Deming's regression analyses comparing each of the test results with LC-MSMS showed slopes that were between 0.881 and 1.217. The interclass correlation coefficients were between 0.92 and 0.97 for all methods. Compared with the serum T concentrations measured by LC-MSMS, the DPC Immulite results were biased toward lower values (mean difference, -90 +/- 9 ng/dl) whereas the Bayer Centaur data were biased toward higher values (mean difference, +99 +/- 11 ng/dl) over a wide range of serum T levels. At low serum T concentrations (<100 ng/dl or 3.47 nmol/liter), HUMC-RIA overestimated serum T, Ortho Vitros ECi underestimated the serum T concentration, whereas the other two methods (DPC-RIA and Roche Elecsys) showed differences in both directions compared with LC-MSMS. Over 60% of the samples (with T levels within the adult male range) measured by most automated and manual methods were within +/- 20% of those reported by LC-MSMS. These immunoassays are capable of distinguishing eugonadal from hypogonadal males if adult male reference ranges have been established in each individual laboratory. The lack of precision and accuracy, together with bias of the immunoassay methods at low serum T concentrations, suggests that the current methods cannot be used to accurately measure T in females or serum from prepubertal subjects.