Do you have any other tests in the suite that use Isolator? I have run into a similar problem before - where tests run successfully by themselves or in their own fixture, but fail when run in the full suite - and it always turns out that some test in some other fixture has their mocks set up incorrectly. For example, one of the other tests in some other fixture may be doing a MockAll when it needs to only mock one instance, or maybe the other test fixture isn't clearing/verifying mocks after a test so there are residual expectations hanging out. That's why I continually stress to my team that they should only mock what they absolutely need. If you only need a call to a method to be mocked one time, don't set the mocks up to always mock the call. That sort of thing.
Unfortunately, if that's the case, finding the problem is going to be a challenge. I generally will load up the tests in the NUnit GUI tester and test things by namespace, working my way up the tree until I either get to running every test in the suite or find some sub-section of the suite that causes the test in question to fail. After that... well, honestly, it's "exclude a test, recompile, run tests... exclude a test, recompile, run tests..." It's hard.
The Tracer utility might help you out, too. It can help you see what expectations are getting set up in each test so you can figure out which tests in the suite are setting up expectations that aren't getting met.