Short answerOnly cortisone has clear support, and it helps for a few weeks. The other injections have not been shown to beat a dummy injection in the best trials, although some pooled reviews disagree.
Evidence grade
Does not helpNot enough evidenceHelps a littleGood evidence
Only cortisone helps, and only for a few weeks
What this page compares
People with knee arthritis are offered four main kinds of injection. Cortisone is a steroid that calms inflammation. Hyaluronic acid is a gel meant to act like the fluid in a healthy joint. PRP (platelet-rich plasma) is made from your own blood. Stem cell injections use cells taken from your own fat or bone marrow, or from donors. This page compares them using what the major guidelines say, and then sets out the newer research published since.
What the guidelines say
Guidelines are written by panels of doctors and other experts who weigh up all the research available at the time. This page treats them as the starting point. The main ones for knee arthritis are the , the and the Australian . A found the three international guidelines were mostly consistent.
Cortisone. Recommended. The American guideline strongly recommends it, OARSI lists it as an option, and the RACGP says it may be offered for short term pain relief. The RACGP also advises watching for harm from repeated use.
Hyaluronic acid. The guidelines disagree. The American guideline advises against it, the RACGP suggests not offering it, and OARSI lists it as an option for some people.
PRP. Not recommended. The American guideline advises against it and the RACGP could not recommend it for or against.
Stem cell injections. Not recommended. The American guideline and the RACGP both advise against them.
The grade at the top of this page follows these guidelines. Where the newer research below points a different way, it is stated openly rather than used to override them.
Newer evidence since the guidelines
Since 2019 there have been new randomised trials and pooled reviews of trials. These are the highest level of evidence, so they are set out here, including the ones that point in different directions.
Cortisone
A with 842 people compared cortisone with a dummy injection. Cortisone gave a difference big enough to notice at 6 weeks or less, and no difference by 6 months. The authors noted the trials were few and of limited quality. This fits the guidelines, which describe cortisone as short term relief.
Hyaluronic acid
A pooled 57 trials with almost 23,000 people and 18 different injections. In the 11 hyaluronic acid trials, pain was no different from a dummy injection, and serious side effects were more common. The same analysis found that trials at higher risk of bias showed bigger benefits than well run ones, and that most injections were consistent with a placebo effect. Triamcinolone, a cortisone, had the best chance of making a noticeable difference, at 2 to 6 weeks, but this rested on a single trial.
PRP
The evidence on PRP points two ways. The best single trial was the Australian . It gave 288 people either three PRP injections or three salt water injections, and the patients, the people giving the injections and the people measuring the results did not know which was which. After 12 months pain had improved by 2.1 points in the PRP group and 1.8 points in the salt water group on a 10 point scale, and cartilage on MRI was no different. The authors concluded the results did not support PRP for knee arthritis.
A comparing PRP with salt water found a small benefit at 3 and 6 months and none at 12 months, with results that varied a lot between trials. Its authors called the value of PRP debatable. Other pooled reviews rank PRP first. A ranked PRP, and PRP with hyaluronic acid, ahead of cortisone, and a found PRP better than the other injections for up to 6 months. Both pool many small trials, and the 2026 review reported very uneven results between them. Its authors are from a clinic that provides regenerative treatments. Pooled rankings can look better than the single best trial for these reasons.
Stem cell injections
A found stem cell injections made no important difference to pain or function compared with a dummy injection.
None of this newer evidence changes the grade above. If a new guideline or a large, well run trial changes the picture, this page will be updated.
Risks and effort
All injections carry a small risk of infection and a sore knee afterwards. The found more serious side effects with hyaluronic acid than with a dummy injection. PRP needs a blood sample and often several visits. Ask what any treatment costs, and what it is likely to add, before you book.
What usually works instead
The guidelines put exercise, education and weight loss (if you are carrying extra weight) first for knee arthritis, ahead of any injection. They also recommend anti-inflammatory gels or creams applied to the knee. Injections are an extra option for flare-ups, not a replacement for those.
Questions people ask
Which injection is best?
For a flare of knee pain, cortisone is the only one the guidelines clearly support, and its effect lasts weeks rather than months. For hyaluronic acid, PRP and stem cells, the best trials have not shown they beat a dummy injection, so none can be called best.
Why do some clinics offer PRP if the guidelines do not recommend it?
Some pooled reviews show a benefit, and many people feel better after any injection, including a salt water one. In the people in both groups improved. That is why blinded trials matter.
How long does a cortisone injection last?
The found the benefit was noticeable at 6 weeks or less and gone by 6 months.
Will an injection stop my arthritis getting worse?
No injection has been shown to do this. In the , cartilage loss on MRI was no different between PRP and salt water.
Sources
Select a source to read what the study or guideline found, in plain words, with a link to the original.
Last reviewed 24 October 2026. Next review due October 2027. General information only. It does not replace advice from your own doctor.
