The short version
- It suppresses inflammation. It does not repair anything. Nothing about the joint is structurally different afterwards.
- Relief is measured in weeks, not months. Pooled trial data show a clear decay, with no detectable effect by six months.
- Repeated high-frequency use has a real question over it — a two-year trial of quarterly injections found more cartilage loss and no pain benefit.
- None of that makes it a bad treatment. It makes it a treatment with a specific job, best used deliberately rather than on a schedule.
Cortisone injections get discussed in two registers. Clinics describe them as a straightforward fix. A certain kind of internet advice describes them as joint-destroying. The literature supports neither, and the actual picture is more useful to know.
What does a corticosteroid injection do?
It delivers a potent anti-inflammatory drug directly into a joint or around a tendon. Inflammation drives a large share of musculoskeletal pain, and suppressing it locally can reduce that pain substantially while avoiding the systemic exposure of oral steroids.
What it does not do is change the structure of the joint. It does not regrow cartilage, repair a tear, or slow arthritis. An injection buys a window in which the joint hurts less — and what happens in that window is what determines whether it was worth giving.
That framing matters more than any other sentence on this page. An injection given alongside a plan — strengthening, load management, weight, activity change — uses the window. An injection given on its own returns the patient to the same joint, doing the same things, with the same trajectory.
How long does a cortisone shot last?
Shorter than most people expect, and the decline is well characterised. A Cochrane systematic review pooling 27 randomised trials in knee osteoarthritis found benefit that fell steadily with time.
| Time after injection | Measured effect on pain |
|---|---|
| 1–2 weeks | Moderate benefit |
| 4–6 weeks | Small to moderate benefit |
| 13 weeks | Small benefit |
| 26 weeks | No evidence of an effect |
So the honest answer to how long will this last is generally a few weeks to a few months, with the strongest effect early and little detectable by six months. Individual response varies widely around that average — some people get considerably longer, some get very little.
It is also worth saying that the review authors were explicit about the limitations: most included trials were small and methodologically weak, and they concluded that whether the benefit at one to six weeks is clinically important remains unclear. That is a more cautious reading than most patients are given.
Do cortisone shots damage cartilage?
This is the question everyone has now, and it deserves a careful answer rather than a reassuring one.
The trial driving the concern randomised 140 people with symptomatic knee osteoarthritis and ultrasound evidence of synovitis to triamcinolone or saline every twelve weeks for two years. At the end, the steroid group had significantly greater cartilage volume loss — and no significant difference in knee pain.
Read the design before drawing the conclusion
That result describes quarterly injections sustained for two years in patients selected for active synovitis. It is not evidence that a single injection, or an occasional one, causes structural harm — that question was not what the trial asked.
What it does establish is that treating a joint with corticosteroid on a fixed schedule, indefinitely, is not a benign default. The pain benefit did not persist, and the cartilage difference did.
The reasonable position sits between the two extremes. Corticosteroid is a useful tool used deliberately and sparingly. It is a poor long-term strategy used routinely, and the trial above is the clearest evidence for why.
Where I have changed my practice
Early in my career I gave injections more readily than I do now, and largely on request. What changed was noticing how often the same joint came back on the same interval — which is a signal that the injection is substituting for a plan rather than enabling one.
That is a judgment formed over thousands of these rather than a finding from a study. But it aligns with what the evidence above shows: the value is in the window it opens, not in the injection itself.
Who does it help most?
Patterns worth knowing, though none of these are guarantees:
- Inflammatory pain responds better than mechanical pain. A joint that is swollen, warm and worse with rest is a better candidate than one that hurts purely with load.
- Earlier disease tends to respond better than end-stage arthritis, where bone-on-bone mechanics are driving the pain rather than inflammation.
- Bursitis and tendon-adjacent problems often respond well, and in some of those the injection genuinely resolves the episode rather than merely postponing it.
- A specific, time-limited goal — completing a course of therapy, getting through a trip, tolerating a strengthening programme — is the best reason to give one.
What this page does not cover
Three related questions have their own pages, because each deserves more than a paragraph: how often an injection can safely be repeated, what to expect afterwards, including flare and blood-sugar rise, and the situations where an injection is the wrong answer entirely.
Sources
- Jüni P, Hari R, Rutjes AWS, et al. Intra-articular corticosteroid for knee osteoarthritis. Cochrane Database of Systematic Reviews 2015;10:CD005328. Pooled time-course of pain benefit across 27 trials. Retrieved via PubMed.
- McAlindon TE, LaValley MP, Harvey WF, et al. Effect of intra-articular triamcinolone vs saline on knee cartilage volume and pain in patients with knee osteoarthritis: a randomized clinical trial. JAMA 2017;317(19):1967–1975. Two-year trial of quarterly injections. Retrieved via PubMed.
- Anil U, Markus DH, Hurley ET, et al. The efficacy of intra-articular injections in the treatment of knee osteoarthritis: a network meta-analysis of randomized controlled trials. The Knee 2021;32:173–182. Comparative ranking across injectable options. Retrieved via PubMed.