Depends for me. If it's something like a mathematical and scientific topic (Riemann Sums, hyperbolic trigonometry, color force, etc) it's absolutely reliable, though not always informative to a layman. On topics like history, yeah, it's quite reliable in general. On recent politics, I don't trust it because it is too often a "young article" that is prone to less revision for style and neutrality. On the whole, though, it's a pretty reliable source. BUT if you're doing intensive research into something it's often not very useful in the first place, unless that something is "the civil war" or some very large topic. More narrow focuses are often best served by looking at other sources (sometimes the sources cited at the bottom are perfectly adequate). Wikipedia covers a lot of topics very broadly. With the exception of high-profile and usually recent topics in the realms of history and politics and so on, even complex and deep topics get less words than is necessary to completely and fully understand them. After all, Wikipedia is still an encyclopedia, not a specialist reference. Not to say it merely skims all topics, since it does have a lot of good stuff, but if you're doing serious research with a narrow focus I find it often simply does not have a sufficient volume of information.
It's one shining virtue, though, is that reading an article there can give you a very good idea of things to look for in subsequent research.