"Faith" is typically exalted by the religious and ridiculed by the non-religious. A closer look at the notion reveals several different kinds of "faith" to which different attitudes are appropriate.
It's widely held by the non-religious that "faith" simply means "believing what, if you were honest with yourself, you'd know to be false" or, at best, "believing without, or against, evidence". It does have such meanings sometimes, but there are at least two entirely reasonable things that are sometimes meant by "faith".
Two rational kinds of "faith"
Planned obstinacy in the face of planned misgivings. Suppose you're a young idealist and form the intention of taking a very lucrative job so that you can send most of your income to starving children in Africa. You're aware that both aging and working at such a job are likely to expose you to all sorts of (what you consider to be) temptations to be selfish and to justify your selfishness by apparently logical arguments, which (human nature being what it is) will seem more convincing to you than they should because accepting them would be in your self-interest. So you also form the intention of being much more skeptical about such arguments than reason seems to justify, and consistently act on that intention.
This appears to me to be entirely sensible and commendable and, yes, rational, even though it involves a deliberate choice not to follow the dictates of reason at all times: because you know that what seem to you to be the dictates of reason will be consistently wrong.
This is, I think, what's going on -- or at least, what should be going on -- when a Christian persists in being a Christian despite apparent evidence against Christianity. Supposing for the sake of argument that Christianity has once been rationally accepted, it is then sensible to take note of its embedded warnings that temptations to abandon it will come along.
- But it's only reasonable within limits. After all, you might have made a mistake originally, and given strong enough evidence the other way you ought at least to rethink, taking the utmost care to avoid the temptations you foresaw.
Tentative experimental belief. Now suppose that after years of raking in the money from that job you've become quite thoroughly corrupted, and devoted to self-indulgent pleasures. Someone tells you of a kind of food, or a sexual practice, that you've never tried before and that, frankly, sounds pretty horrible; but he assures you that he thought so too at first but has found it wonderful and life-enhancing. You've generally found him a useful guide to the world of gluttony and lust, and so you suppress your misgivings and give whatever-it-is a try ... and, indeed, it's very enjoyable.
So far, so obvious, and (aside from moral concerns) this is clearly sensible. But what if the thing that's being suggested is something that only works if you're in a particular mental state? (Unlikely for food, but plausible enough for sex.) Let's drop the analogy, and turn to the actual topic at hand: suppose a bunch of generally sensible people, who seem to have happy and well-ordered lives, tell you that they got that way by following a certain religion; and that in order to get the benefits of following it, you need to believe certain propositions. After they've explained what those propositions are, and why they accept them, you aren't fully convinced but they seem plausible; and it does seem that these people have derived great benefits from their religion. If you happen to be psychologically capable of it, it seems pretty reasonable to say: "OK, I'll believe these things, so far as I can, and see what happens.".
- But it's only reasonable within limits. After all, those people might be making some sort of mistake, or there might be some reason why what works for them doesn't work for you. If you try "believing these things" and don't run across any good reason to carry on, you ought to remember that your belief was only provisional and retract it.
"Salvation by faith"
Christianity notoriously goes further and claims that "faith" can be not merely reasonable, but tremendously valuable, and even the key to salvation. I think this is a third meaning; it's more about trusting a person (namely God, or Jesus, or both) than about believing a set of propositions. (Though the latter does occur, as e.g. in Romans 10:9.) Given the assumptions that God exists and that Jesus is his Son (or his representative, or even merely his prophet), it's entirely reasonable to regard trusting God and trusting Jesus as Good Things.
- But it's only reasonable within limits. After all, if it should turn out that there is no God or that Christianity is terribly wrong about what he's like, then your trust would be misplaced. If you encounter enough apparent breaches of the trust you've placed in him, or good enough evidence that he doesn't exist or isn't trustworthy, then you should stop trusting him.
Belief contrary to reason
None the less, the term "faith" is sometimes used to mean what the non-religious take it to mean. When it does, the sort of "faith" being described is positively irrational. Likewise when any of the foregoing kinds of "faith" is carried too far, as I've pointed out case by case above.
Faith as a "way of knowing"
Sometimes people say things like this:
Religion is a way of knowing that through revelation, practice, and faith, comes to conclusions about the supernatural world.
It's not clear exactly what role is envisaged for faith here, but if the idea is that faith constitutes a way of finding things out, I think it's all wrong. Faith is sometimes a sensible attitude to take to something you already know in some other way, but if you say "I know this by faith" as if it's a justification for your belief then you owe the world an explanation of why your faith in something constitutes a justification of it.
Such statements are sometimes abbreviations for statements that make sense, like "I know this because I was told it by someone I trust" (where the "someone" might be, say, the Roman Catholic Church or the Bible). Then, of course, the question arises: why do you trust that source? So, in these cases, "faith" is once again derivative, based on a pre-existing trust in someone or something else. I think this is true of all cases where "faith" is reasonable.