It's hard to hide severe back pain. When I stand up, I look like a question mark. The visibility of the problem, combined with the general goodness of my fellow human beings, leads to lots of unsolicited advice. Folks have given me great advice (take some NSAIDs, stretch, and don't lay in bed) and some questionable advice (go to the chiropractor, get some acupuncture). My colleagues and I have written a lot about acupuncture. It's sort of a "gateway CAM", in that it has a patina of plausibility. But the evidence of its efficacy has pointed toward it being an elaborate placebo.
To help settle the question, the British Medical Journal (BMJ) has published a systematic review (which is different from a meta-analysis) of studies of acupuncture that were as well-controlled as possible. Much of the acupuncture literature suffers from lack of proper controls, as it's difficult to blind either the subject or practitioner to pointy needles in flesh. But it can be done, to a certain extent, with "sham" acupuncture, and the current review aimed, "to analyse all trials of acupuncture for pain that had two control groups consisting of placebo acupuncture and no acupuncture." This type of systematic is especially important for implausible medical claims. If you do enough studies, you can always find "positive" results---a key question is do the positive results reflect reality. A well-done systematic review can help determine what the preponderance of the data mean.
So, what did they find?
It turns out that so-called placebo acupuncture has some analgesic effect compared to no acupuncture at all, but that both "real" and "placebo" acupuncture provide no significant benefit that can be separated out from the effect of undergoing an intervention.
In other words, if someone is in pain, and you do something to them, they will often report feeling better. Acupuncture shows no consistent effect above and beyond this.
This should settle the issue. It seems like there's a new acupuncture trial every couple of months, and none has been particularly ground-breaking. This systematic analysis should be the end of it. No matter how many times you try it, it's still, in aggregate, no better than placebo. This study, combined with the greater part of the available evidence, should lead us to abandon support for acupuncture.
M. V. Madsen, P. C Gotzsche, A. Hrobjartsson (2009). Acupuncture treatment for pain: systematic review of randomised clinical trials with acupuncture, placebo acupuncture, and no acupuncture groups BMJ, 338 (jan27 2) DOI: 10.1136/bmj.a3115