but are getting quite widespread reactivity even in control brain.
This is a major issue in apoptosis research. There are plenty of scientific papers which show such kind of staining in otherwise normal nuclei and call these cells apoptotic cells. I think the staining of normal nuclei is real and not background or an artefact, however, real apoptotic cells in the CNS show caspase-3 in the cytoplasm. Most of the time the nuclei of these cells reveal condensation, but not always ( depending on the form of apoptosis). We have always used the CM1 antibody which we got non-commercially from IDUN. For a certain time this antibody was sold by BD Pharmingen, but not anymore. BD Pharmingen however sells other activated caspase-3 antibodies which in our hands give the same results. I think however that your antibody is working perfectly. We occasionally also see such (formalin-dependent) nuclear staining with our caspase-3 antibodies (especially in rat and mice). If this nuclear staining is very strong you might dilute your antibody some more. Real apoptotic cells then will still be strongly positive. You can check out my recent for examples of apoptotic cells in the brain:
Bauer J, Elger CE, Hans VH, Schramm J, Urbach H, Lassmann H, Bien CG. Astrocytes are a specific immunological target in Rasmussen's encephalitis. Ann Neurol. 2007 Jul;62(1):67-80.).