Permeability of crystalline rocks results mainly of cracks. Using percolation theory and a statistical approach, Dienes has recently derived a model which appears to be very adequate to describe and calculate permeability in such rocks at a scale which is larger than the scale of the cracks. Using this model, we show that three microstructural parameters are of primary importance : the crack shape factor A, the mean crack length and the average crack spacingAlthough the third one, is frequently presented as the most important parameter, we suggest that A andare indeed as much important. Percolation transition depends on andbut high permeabilities are obtained if either A or are large. Using these results we discuss time constants for permeability evolution by taking into account slow crack growth processes. We suggest that slow crack growth could result in large increases of permeability with or without developing a macroscopic fracture. Such effects are of interest for seismogenesis, underground repositories and other problems concerned with fluid flow in the crust.