When using Rev Thrust the possibility of weathercocking in crosswinds on a slippery must be considered.
If you aply reverse in calm conditions the thrust vector is applied along the longtitudinal axis of the a/c and the retarding force is therefore at its most efficient.
However, if there is a crosswind, and if the wheels do not have proper grip - the a/c may weathercock into wind. If this happens, the reverse thrust effect will no longer act in the same way. Instead there will be a split in the applied vector, part will act to retard the a/c, but a part will also act to 'slide' the a/c in the downwind (off RWY) direction.
Effectively the reverse thrust acts to drag the a/c off the RWY.
The only cure is to reduce the applied rev thrust and get back on c/l...then re-apply rev.
A diagram would best describe this situation...perhaps someone can provide a link.