Great info. Thanks to both. I'll mull it over for a bit before I decide.
Something else to keep in mind: Beyond about 15A of current needs, it is generally better to buy than to build, simply because to design at those levels, you need a very good understanding of how to design such h-bridges to avoid burning up your components. Unless you are attempting to gain such experience (vs just wanting an h-bridge that will "do the job"), it can end up being a very expensive endeavour (particularly if you are going the "discrete" component route - burning up $15-20.00 USD MOSFETs can be an expensive learning experience).
At a certain point, you need to weigh the component costs vs a pre-built board, and whether you are willing to burn up a few; you can still smoke a pre-built board, but generally if you take the time to research your needs and the specs of the devices you are hooking up, and hook up everything properly (and program for it properly), pre-built motor drivers are fairly rugged and forgiving.
For very large motors (>25A or so) - pre-built is likely the best way to go; building your own driver for this size is possible, but you likely won't save any money (and you may even spend more) - and your driver may or may not be any better in performance (depending on your design skill, of course).
For small projects (like a robot platform made using a small R/C tank chassis or similar), using an L298 or other type of small driver (even a small discrete component h-bridge) is probably the cheaper method, especially if you can find the components for the bridge surplus. It can also be a great learning experience that, if you wanted to, could be used as a good foundation to design and build larger h-bridges...
