FWIW, the Cochrane Library did a systematic review of the literature in 2009 of RCTs that compared Glucosamine (only) vs placebo. They found 25 trials that met their criteria.
I will give findings in standard mean difference (SMD) scores. The more negative the score, the more in favor of glucosamine, the more positive, the more it favors placebo. If it crosses 0, it is usually regarded as not statistically significant, or clinically significant.
Pain; glucosamine vs placebo: SMD = -0.47 (-0.72 to -0.23).
Lequesne score; glucoasamine vs placebo (an OA severity score): SMD of -0.47 (-0.82 to -0.12)
Glucosamine vs NSAIDs: SMD of -0.27 (-0.65 to 0.11) — crosses 0.
There are several other findings in the study, most were about toxicity and others that weren't pertinent to this discussion. And most crossed 0.
FYI, the cochrane library is in the middle of a systematic review of chondroitin vs placebo.
I agree with others that it works for some and not for others. I speculate this is due to the individual metabolism of each person's ability to break down the compounds. This is huge area of study in medicine as some people respond to medicine better than others. Usually due to differences in metabolic enzymes. There are actually certain places that type your CYP enzymes before using general anesthesia to attempt to reduce prolonged anesthesia due to difference in enzyme isotypes.
Cheers,
Josh