The detection of testosterone and its precursors' abuse in antidoping sports analysis is based on the longitudinal evaluation of markers of the urinary endogenous steroid profile. A Bayesian statistical approach is applied, allowing the establishment of credible intervals of the selected parameters for every athlete. Samples showing values outside the acceptable boundaries are selected for additional confirmation by isotope ratio mass spectrometric (IRMS) analysis. The alterations of the IRMS values last longer than the alterations of the steroid profile. Then the application of IRMS to a larger number of samples, at a screening level, would presumably allow detection of additional positive cases. The steroid profile and IRMS data can be treated using the same Bayesian inference procedure. In nonsports population, we have demonstrated the stability of IRMS data. In this work, we studied the variability of these data in real conditions, in samples collected on athletes subjected to antidoping analyses over the years. The data obtained confirmed previous observations and the applicability of the proposed approach. The results of cases where confounding factors of the steroid profile were reported are discussed, showing that in most of the cases no significant changes are observed over the absolute delta values. Changes in diet may significantly change the absolute delta values but not the ones relative to endogenous reference compounds. Finally, a case that could have been evaluated as normal with the current approach without a thorough review of the data was detected as positive by the proposed approach, demonstrating the benefit of its application.