Metabolic stability is a key issue in the development of orally active androgens for Partial Androgen Deficiency in Aging Males (PADAM) and male contraception. Rates of metabolism in human hepatocyte suspensions provide useful information on the stability of compounds that undergo a first pass metabolism. We have derived a structure-pharmacokinetic relationship for a data set of 32 in-house steroidal androgens by means of the decision-trees technique. Volume, shape, number of rotatable bonds, and surface turned out to be the most important descriptors for classification. Only 2 of the 32 compounds were misclassified. The most stable compounds were classified in three leaf nodes on different branches of the tree, suggesting that higher metabolic stability can be achieved for the same substrate by different steric modifications. Further, it is generally assumed that the first step in cytochrome P450s oxidation reactions takes place by hydrogen abstraction to form a radical intermediate. An electronic model for hydrogen abstraction in steroidal androgens was, therefore, developed by means of ab initio calculations. Activation energies of steroid radical systems calculated as energy differences between the reactants equilibrium geometry energies and their corresponding transition states energies could be used to predict relative rates of metabolism to guide the design and redesign process of metabolically more stable steroidal androgens.