When researchers set out to test whether two nutritional supplements helped relieve arthritis pain, many were hopeful that a clear-cut answer would emerge about the popular yet unproven alternative treatment.

Previous smaller studies suggested the supplements — glucosamine and chondroitin sulfate — were effective at treating aching joints. But many of the studies were flawed or paid for by the supplement makers.

The latest study, funded by the National Institutes of Health and published in Thursday's New England Journal of Medicine, found the supplements overall did little to ease osteoarthritis, the most common form of arthritis.

But patients who had more severe knee throbbing seemed to show some improvement.

"We still have a bit of a conundrum," said Dr. Tim McAlindon, a Tufts University rheumatologist who had no role in the research.

Osteoarthritis afflicts more than 20 million Americans and that number is expected to double in the next two decades as baby boomers age. Osteoarthritis is a degenerative joint disease that affects the knees, hips, back and the small joints in the fingers.

Based on the findings, people with severe arthritis should talk to their doctors about trying the supplements short-term to see if they work, said rheumatologist Dr. Daniel Clegg of the University of Utah, who led the study.

Worldwide sales of glucosamine and chondroitin topped $1.7 billion last year, according to the Nutrition Business Journal, which tracks supplements.

The supplements — made from animal cartilage and shellfish — have had even wider appeal amid safety concerns over certain painkillers, including the blockbuster arthritis drug Vioxx, which was pulled from the market in 2004.

At least 5 million Americans use the two supplements either alone or together, government figures show. President Bush used to take the supplements to relieve stiffness in his joints, but no longer uses them.

The supplements showed no serious side effects during the government's six-month study, considered the largest and most rigorous test of the two supplements to date. But the scientists didn't address the safety of longer-term use.

The arthritis research is the third major study in a year to find no overall benefit from some of the most popular nutritional supplements. Recently, research showed the herb saw palmetto didn't reduce symptoms of an enlarged prostate, and last year a study indicated echinacea didn't prevent or treat colds.

Unlike drugs, such supplements are loosely regulated, and their makers don't have to prove the products are safe or effective.

Irene Schwartzburt, a retired teacher from Plainview, N.Y., said she plans to keep using the supplements. The 72-year-old said the remedies relieved the "sticking pain" in her right knee when painkillers failed.

"I want to stay active," she said. "The supplements work for me so why not continue with them?"

In the government study, 1,583 patients with arthritis knee pain received one of five treatments: either glucosamine or chondroitin, a combination of both, the painkiller Celebrex or dummy pills. Neither the doctors nor patients knew which treatment was given.

After six months, patients filled out a questionnaire to determine how many felt a 20 percent reduction in pain. Researchers found the supplements when taken alone or together were no more effective than dummy pills at pain relief.

Sixty percent who took the dummy medication had reduced pain compared with 64 percent who took glucosamine, 65 percent who took chondroitin and 67 percent who took the combo pills. These differences were so small that they could have occurred by chance alone.

The drug Celebrex did reduce pain — 70 percent reported improvement — affirming the study's validity. However, the drug is being studied to see if it's safe for people at risk of heart problems.

Of the 354 people with moderate to severe pain, 79 percent who took both supplements reported relief compared with 54 percent who took the dummy pills and 69 percent who took Celebrex.

In a journal editorial, Dr. Marc Hochberg of the University of Maryland noted the study's limitations: a high dropout rate (20 percent) and a whopping 60 percent who said the dummy pills made them feel better — double the usual placebo effect. Hochberg has received consulting fees from Pfizer Inc., which makes Celebrex, and Merck & Co., which made Vioxx.

Clegg and 10 other researchers in the study reported receiving fees or grant support from Pfizer or McNeil Consumer & Specialty Pharmaceuticals, which makes Tylenol.

The Council for Responsible Nutrition, which represents dietary supplement makers, said it was pleased about the positive findings in the severe arthritis group.