Five sets of testosterone derivatives have been studied separately in respect of testosteronereceptor interaction and in the development of QSAR models. Derivatives of testosterone have been divided in five sets on the basis of the method of measurement of their biological activity. The study has been made with the help of eight topological descriptors; solvent accessibility surface area (SASA), molar refractivity, valence connectivity index (of order 0, 1 & 2) and shape index (of order 1, 2 & 3). The values of descriptors have been evaluated with the help of CAChe Pro software. QSAR models of the derivatives of first, fourth and fifth sets have high degree of predictive power as the value of correlation coefficient (r2) is greater than 0.8. The molar refractivity, solvent accessibility surface area (SASA) and shape index (of order 1) are the best descriptors for QSAR models. The results indicate that London dispersive forces appear to play an important role in testosterone-receptor amino acid interaction. Solvent accessibility surface area (SASA) of testosterone is useful in the measurement of extent of interaction of testosterone with environment.