There are several limiting factors when OC a video card...... The first one, of course, is heat where the load temps are starting to reach the 90c's/100c's range which could potential damage the card. Usually aftermarket coolers solve this problem as well as turning the fan up on the stock cooler.
Next limiting factor, for some people, is just overall stability and/or that the card just isn't meant for OC.... I have found that some mobo's can determine your OC capabilities/stability, ex. OC my x1600 Pro on my Compaq would sometimes lead to a crash or OC options were unavailable with certain programs while my new mobo had no problems........
The next limiting factor is/are artifacts which, how should i put it, graphic anamoles, usually in the forms of triangles, that signify that your card can't keep up with the demands of the OC and could potential damage the card after prolong periods of the OC....... OC programs usually/should come with a 'scan for artifacts' utility that will check for artifacts.... Usually one or two artifacts aren't a big deal but that should signal your limit to your OC.......
If you're are able to, upping the voltage can lead to more stable OC with less/ no artifacts at the price of a hotter GPU...... Some mobo/utilities allow you to do this although upping the voltage could potential shorten the life span of the video card......
Usually with video cards, it's recommended that you go up 5-10 Mhz at a time, OC only the GPU or the RAM at a time, then testing for stability/artifacts each time...... Although, if you already know your clock speed limits, you can OC straight to the specific clock.... Although this should be a slow, progressive process with checking overall stability each time, all you really have to do is check temps and artifacts at each clock before it's usually safe to OC some more and it usually take a half an hour to an hour to find your max clocks......