In a previous paper, we have demonstrated the importance to define a statistical model describing the observed linear correlation between the absolute magnitude $M$ and the log line width distance indicator $p$ of galaxies (the Tully-Fisher relation). As long as the same statistical model is used during the calibration step of the relation and the step of the determination of the distances of galaxies, standard statistical methods such as the maximum likelihood technic permits us to derive bias free estimators of the distances of galaxies. However in practice, it is convenient to use a different statistical model for calibrating the Tully-Fisher relation (because of its robustness, the Inverse Tully-Fisher relation is prefered during this step) and for determining the distances of galaxies (the Direct Tully-Fisher relation is more accurate and robust in this case). Herein, we establish a correspondence between the Inverse and the Direct Tully-Fisher approaches. Assuming a gaussian luminosity function, we prove that the ITF and DTF models are in fact mathematically equivalent (i.e. they describe the same physical data distribution in the TF diagram). It thus turns out that as long as the calibration parameters are obtained for a given model, we can deduce the corresponding parameters of the other model. We present these formulae of correspondence and discuss their validitity for non-gaussian luminosity functions.