The current approach to detection of doping with testosterone is based on measuring the testosterone to epitestosterone ratio (T/E) in urine by gas chromatography/mass spectrometry. The median T/E for healthy males who have not used T is about 1.0. In a single urine, a T/E lower than six leads to a negative report even though it does not exclude T administration. A value greater than six indicates possible T administration or a naturally elevated ratio. It has been shown previously that the carbon isotope ratio of urinary T changes after T administration. In this study a potential confirmation method for T abuse was optimized. Gas chromatography/combustion/carbon isotope ratio mass spectrometry (GC/C/IRMS) was used to analyze two T precursors (cholesterol and 5-androsten-3 beta, 17 beta-diol) and two T metabolites (5 alpha- and 5 beta-androstane-3 alpha, 17 beta-diol) in addition to T itself in each of 25 blind urines collected from eight healthy men before, during or after T administration. The carbon isotope ratios of T and the metabolites were lower after T administration. The relationships among the variables were studied using multivariate analysis and beginning with principal components analysis; cluster analysis revealed that the data are composed of two clusters, and classified the samples obtained after T administration in one cluster and the remainder in the other; discriminant analysis correctly identified T users. The measurement of carbon isotope ratios of urinary androgens is comparable to the T/E > 6 test and continues to show promise for resolving cases where doping with T is suspected.