The reactivation kinetics of hydrogen‐passivated Mg acceptors in InP have been studied by annealing experiments carried out with different reverse biases. It is shown using a new analysis that the actual dissociation energy of the Mg‐H complexes can be estimated even without applying a sufficient reverse bias to overcome retrapping of H at the dopant site. The dissociation process follows a first‐order kinetics and the dissociation frequency and activation energy were estimated at various depths ranging from the surface up to 0.5 &mgr;m using an empirical analysis of the experimental data. A bias‐independent dissociation energy of 1.40±0.08 eV was obtained when estimated at the surface. An apparently higher dissociation energy results when calculated in the bulk. This overestimation is pronounced for low‐bias anneals and is explained as a result of retrapping. The concentration profile data is consistent with the positively charged state of the diffusing H inp‐InP.