Good Statistic For Blocking

We are having an occasional problem occur where a
process will not stop blocking.

We are trying to trace the problem, but in the interim,
I would like to set up an alert that notifies me when a
process has been blocking for too long.

Are any of the lock wait times good statistics to use
for such an alert? If not, is there anything else I could
look at from the alert level?

If I had to, I could periodically create a table of
sysprocess spids that are at the top of blocking chains,
then test for a spid that lingers. I'm hoping I can
avoid this and use the built-in monitoring instead, though.

Thanks!
Geoff