Whenever I am monkeying with core voltage, temperature, multiplier, fsb, etc, I use P95 Torture as the first test. If the processor makes it out of the first pass of Test 1, it is probably not going to be the limiting factor, but then, by the end of that first pass, on my system, the temp has stabilized. On those rare occasions where it has not stabilized, I let it run a second pass.
Since your temperatures are not out of line, I'd say up the core voltage another .025v and try again, assuming you are not maxed out. I can tell you that my Duron 600 which runs 10x100 reliably under P95 at 35C and 1.8V will not if I allow the core temp to go up to 45C. Somewhere around 42C, I have to bump up the core voltage another step. Be careful of falling into the higher temp needs higher voltage, makes higher temp, needs higher voltage... Spiral.
Hoot