First off, get a program that can monitor your GPU's temperature, such as Rivatuner.
This thread (on another site) might be of use, or just have a look around on Google. I'm not sure of the best tool available (I've never really had a need for using anything like that myself), so maybe someone else can help in that field of knowledge (and my lack of it).
Generally, when your GPU is under load, you'll want it to keep it under 100c. And that's the
maximum temperature you should have it at. Ideally, you'll want cooler temps around, say, 50-80c is okay. As always, cooler is always better.
Now, if the GPU temp remains fairly cool (eg, under 70-80c or so), it's most likely not the problem. Still, if it's above 60c, you may want to consider a 3rd party cooler for your GPU. It's not essential, but it may be a good idea, and it's probably a bit quieter too. Zalman make several good models, and so do Arctic Cooling. Again, Google can help you.
So, if the GPU isn't the problem, it may be the PSU. Like you said, if it's getting hot, it could be several things...
1) Dust buildup (make sure your case is as dust free as possible);
2) Not enough cooling (consider adding more fans to your case if possible, and maybe tidying up the internal cabling to allow for more airflow); or
3) The PSU is underpowered. Ideally, if you have a fairly modern machine with, say, a Core 2 CPU, and a mid to high end graphics card, you'll want a fairly good power supply. For a machine of that spec, I generally stick to about 500W or more.
Those are some basic steps you can take to check for the source of the problem, so try those first.