Doping control, or testing, is one of the essential components adopted by anti-doping organizations to protect clean sport competitions. Alongside the evaluation of athletes’ samples for prohibited substances or methods, the Athlete Biological Passport (ABP) has been established, for some specific disciplines, as a complementary pillar in the detection of doping, since its introduction in 2009. The fundamental principle of the ABP is to monitor over time athletes’ individual profiles – with respect to certain doping biomarkers such as testosterone – that may reveal anti-doping rule violations. Significant variations from an athlete’s established levels can be further assessed for possible manipulation. Currently, the practical implementation of the ABP framework is based on a Bayesian approach called ADAPTIVE. Specifically, given a biomarker, the predictive posterior distribution of a future sample is used to determine individual limits discriminating between normal and anomalous values. These individual limits are then continuously updated as additional samples are taken, and the observed values are compared against them to identify potential anomalies. However, the ADAPTIVE approach is implemented on longitudinal profiles of single markers (e.g., testosterone), of single combinations of two markers (e.g., testosterone over epitestosterone; the so-called T/E ratio), or, alternatively, of a few biomarkers following a univariate approach. Inspired by the use of copulas for modeling multivariate datasets, in this work we extend the established ADAPTIVE method to multivariate testing of longitudinal ABP profiles. We express the multivariate joint distribution of a set of biomarkers through a separate modeling of the marginal distributions and of their dependence structure. Focusing on a parametric setting, we evaluate the performance of such a multivariate approach in a number of simulation studies, varying according to the copula family and the number of biomarkers. Joint work with Brunero Liseo