Permeability estimation of poor permeable formations like tight or gas-shale reservoirs using a pulse-decay experiment performed on crushed samples has been shown in earlier works to be an interesting alternative for it is faster and less expensive than traditional transient tests performed on carefully prepared core plugs, although it is restricted to measurement in the absence of overburden pressure. Due to reservoir depletion during production, sample characterization over a wide range of pore-fluid pressure is essential. If the Darcy-Klinkenberg model is thought to be a satisfactory gas-flow model for these tight formations, the full characterization can be achieved by determining both the intrinsic permeability, kl , and Klinkenberg coefficient, b. In this work, the conditions under which reliable estimates of kl , b and porosity, φ can be expected from this type of measurement are carefully analyzed. Considering a bed of monodisperse-packed spheres and a complete physical model to carry out direct simulations and inversion of the pressure decay, important conclusions are drawn opening wide perspectives for significant operational improvement of the method. In particular, it is shown that: i) The particle size of the crushed sample must be well selected for a reliable pressure-decay signal record. ii) The simultaneous determination of both kl and b by inversion of the pressure-decay signal is very difficult because the sensitivities of the pressure decay to both coefficients are correlated. iii) The porosity of the particles can be accurately estimated when the experimental setup has been properly calibrated (volumes of the chambers and of the porous sample). The precision on the estimation of this parameter is however strongly dependent on a bias on the crushed sample volume. iv) When identification of kl and b is possible, a very significant error may occur in the determination of the intrinsic permeability due to a bias on the porous sample volume. Errors on the estimated values of φ and kl due to a bias on the chamber volume are not very significant. Moreover, b remains insensitive to bias on both the chamber and porous sample volumes.