The function points pertain to complexity that in turn provides input into the probable effort that you require to build the code.
Test case points pertain to the risk exposure for different functions. The risk exposure does not have to map to complexity (it is often true that more complex code is higher risk code, but don't fall for function points as a test priority code).
Usually you look for user contact, loss of data, liability, and that sort of exposure to estimate the relative consequences of failurs of different aspects of functionality. You then apply that so you emphasize the testing of code where you can least afford a failure during execution.
I typically use a "FEA" score (derived from FMEA scoring in 6-Sigma) where I estimate the exposure to Failure, the possible Effect, and opportunity for Avoidance, and rate them 1 (high priority) to 5 (low priority). By multiplying F * E * A you do get a relative priority 1 - 125 that you can use to rank your testing effort. Nothing scientific, just a yardstick (I have had my hands slapped not to make it sound too formal).
This is the flipside of a number of posts related to the disappearance of jobs to offshore testing services: the fact that competitive bids may be offered when there is really not enough to go on. This is a business risk, and what many people are telling us is that low bidders may end up cutting scope in order to make a profit despite the lower bid. Clearly, you cannot properly estimate unless you know what you are supposed to test, all you can do is gamble that you ask for enough money. I believe that asking for that information prior to submitting a bid is not a bad move, and it would also protect the client from a situation where a vendor has no choice but to cut corners in order to make a profit on a low bid. You may lose some bids (we are getting used to that over here) but you will at least save your reputation. I still think there is enough business out there that you can do it right and still make money, while your client gets the benefits. After all, a project at any price that does not deliver the goods is one for which the client overpays.