Dislocations in ice behave very differently from those in other materials due to the very low energies of stacking faults in the ice basal plane. As a result, the dislocations dissociate on the basal plane, from a perfect dislocation into two partial dislocations with equilibrium width we ranging from 20 to 500 nm, but what is the timescale to reach this dissociated state? Using physical models, we estimate this timescale by calculating two time-constants: the dissociation-completing time td and the dissociation-beginning time tb. These time constants are calculated for two Burgers vectors as a function of temperature. For perfect dislocations with Burgers vector <c + a>, td is more than one month even at the melting temperature TM, and it exceeds 103 years below −50 ℃, meaning that the dissociation cannot be completed during deformation over laboratory timescales. However, in this case the beginning time tb is less than one second at TM, and it is within several tens of minutes above −50 ℃. These dislocations can glide on non-basal planes until they turn to the dissociated state during deformation, finally resulting in sessile extended dislocations of various widths approaching to the equilibrium value we. In contrast, for perfect dislocations with Burgers vector <a>, td is less than one second above −50 ℃, resulting in glissile extended dislocations with the equilibrium width we on the basal plane. This width is sensitive to the shear stress τ exerted normal to the dislocation line, leading to extension of the intervening stacking fault across the entire crystal grain under commonly accessible stresses. Also, due to the widely dissociated state, dislocations <a> cannot cross-slip to non-basal planes. Such behavior of extended dislocations in ice are notable when compared to those of other materials.