I agree with Marc Alspector-Kelly (via Undetached Rabbit Parts) that it's useful to separate the internalism/externalism debate about justification, from the parallel debate about knowledge. It sounds like these are often run together (since people often assume that 'justification' is a conceptual component of 'knowledge'), but I can't go along with that because - like Marc - I'm an internalist about justification but prefer externalism about knowledge.
I take it that justification relates to what we (epistemically) ought to believe, based on the evidence available to us. When put like that, it's practically internalist by definition. Externalism claims that what matters is (partly) external to the agent - not accessible to introspection. But if such properties are not internally "available to us", then we cannot use them as evidence when judging what to believe. So they can play no role in justifying our beliefs. [Though cf. John at FBC.] If they have importance (as they do), it is for something besides justification.
That 'something', I suggest, is knowledge. Internalism doesn't work for knowledge, because knowledge requires a more objective 'truth-connection', that mere internal justification may not provide (see, e.g., Gettier cases). Of course, truth itself is another component of knowledge, but I don't just mean that. Rather, our reasons for belief must be somehow tied to the truth. I won't go into any detail here though; see Marc's paper for a better explanation.
(By the way, my favourite account of knowledge is Keith DeRose's possible-worlds externalism, which I mention in my skepticism overview, and explain in more detail here. For justification, however, I favour coherentism - a belief is justified to the extent that it coheres with our other beliefs.)
Perhaps some epistemologists use the word "justification" stipulatively to simply mean "whatever it is that must be added to true belief to yield knowledge". But then, they're not really talking about justification at all, but some other concept altogether - namely, whatever knowledge has over true belief.
So, I'm wondering, is there any reason not to separate the two debates as I (following Marc) have done here?