When held at temperatures above the glass transition temperature, crazes in certain polymers may be made to heal; that is, the crazed material may be made to recover the mechanical properties it had prior to crazing. Using thin films of a variety of polymers, we have investigated whether the mechanism of entanglement loss during crazing influences the heat treatment time necessary for healing. We find that under experimental conditions for which there is evidence of disentanglement during crazing, and if the crazes do not break down, healing occurs after heat treatment times of the order of those necessary to make the craze disappear optically. Similar heat treatments applied to scission mediated crazes however, do not result in healing. We argue that scission crazing results in a high proportion of chain fragments which are unable to contribute to the entanglement network. Since the heat treatments are not sufficiently long to disperse these fragments, and hence to restore the original local entanglement density, healing does not take place. Disentanglement should not result in such damage, so, consistent with our observations, healing times should be relatively short. Hence, these results provide independent evidence for disentanglement during crazing.