Every surgeon has a default graft, and every graft has a track record. This page sets out what the registries, the randomised trials and the meta-analyses show about each, including the donor-tendon grafts this practice uses often, with the evidence on both sides, so that the choice made for you is one you understand.
The four grafts in one table
| Graft | Strength of evidence for durability | Donor-site cost | Best suited to |
|---|---|---|---|
| Hamstring (your own) | Good; about twice the revision risk of patellar tendon in registries; worse when the graft is 8 mm or less | Hamstring weakness for a year; back-of-thigh ache; numb patch on the shin | Most adults with adequate tendon size; people who kneel for work |
| Patellar tendon (your own) | Best in registries; most stable on examination | Front-of-knee pain; kneeling pain in a proportion; small loss of extension | Professional and high-level competitive athletes |
| Quadriceps tendon (your own) | Matches patellar tendon for failure in meta-analyses; newer, thinner data; a registry signal of more re-operation for stiffness | About a quarter of the patellar tendon’s donor-site pain; slower quadriceps recovery | The high-level athlete who wants to avoid kneeling pain |
| Donor tendon, non-irradiated | Equal to hamstring in the largest registry and in meta-analyses of non-irradiated tissue; higher failure in young high-demand patients and with any irradiated or chemically processed tissue | None | Most patients at any age, including recreational athletes; small hamstrings; revision and multi-ligament surgery |
Hamstring versus patellar tendon
The two grafts with the longest record. The Norwegian registry, 12,643 reconstructions, found a 5-year revision rate of 4.2% overall and a hazard ratio of 2.3 for hamstring against patellar tendon, highest in the youngest patients. The combined Scandinavian registries, 45,998 reconstructions, put the hazard ratio at 0.63 in favour of the patellar tendon, with the difference shrinking with age. In patients 25 and under, a 2026 meta-analysis of 46 studies found failure of 11.1% with hamstring against 5.1% with patellar tendon.
The Cochrane review explains the trade. Patellar tendon reconstructions were more stable on every test, and patients had more pain at the front of the knee and with kneeling, a small loss of extension, and a trend to weaker quadriceps. The 2015 meta-analysis of 22 studies found no difference in laxity or failure, a better pivot-shift test and return to pre-injury activity with patellar tendon, and less anterior knee pain with hamstring.
Hamstring size
The hamstring graft’s weakness is its variability. The tendon you were born with sets the graft’s diameter, and diameter predicts failure. Across four studies, a graft of 8 mm or less carried a 6.8 times relative risk of failure. In one series, revision was 1.7% with grafts over 8 mm, 6.5% at 7.5 to 8 mm and 13.6% at 7 mm or less, and 14.3% in patients under 20 against 0.7% in those 20 and over. In the Kaiser registry, each extra half millimetre between 7 and 9 mm lowered the revision risk by 18%. Tendon size can be estimated before surgery from the MRI and from height and build, and a small predicted graft is one of the reasons the hamstring is not the default here: a donor tendon sidesteps the problem entirely, a patellar tendon does so for the high-level athlete, and a hamstring augmented with donor tissue, which in pooled data fails no more often than hamstring alone, is a third option.
Quadriceps tendon
In the 2019 meta-analysis of 27 studies, the quadriceps tendon matched the patellar tendon for laxity and failure with about a quarter of the donor-site pain, and matched hamstring for failure with slightly better knee scores. A 2024 meta-analysis found about half the failure risk of hamstring, and the 2026 pooled data in patients 25 and under put its failure at 2.5%. Against that, a Swedish registry of 5,653 cases reported a re-operation rate of 19.2% with quadriceps tendon against 8.9% with hamstring and 6.9% with patellar tendon, mostly for stiffness and pain, and the thigh muscle is slower to switch on afterwards. It is a good graft whose long-term picture is still forming.
Non-irradiated donor tendon: the evidence on both sides
A donor tendon, an allograft, is tendon from a tissue donor, screened, prepared and frozen by a tissue bank. It has no donor site: no hamstring to weaken, no patellar tendon to make kneeling painful, no second incision, a shorter operation and an easier first few weeks. The question has always been whether it lasts.
Processing is most of the answer. Allografts can be sterilised by irradiation or by chemical treatment, and both weaken the tissue. In the Kaiser Permanente registry’s analysis of 5,968 allograft reconstructions, grafts irradiated above 1.8 Mrad carried 1.6 times the revision risk and chemically processed grafts 2.5 times, compared with other preparation methods. In the same registry’s comparison of 14,015 soft-tissue grafts, allografts irradiated at 1.8 Mrad or more carried 3.4 to 3.9 times the revision risk of hamstring autograft, and chemically processed grafts 3.0 times, while non-processed allografts and those with minimal irradiation were not found to have a different risk of revision from hamstring autografts. Much of the older evidence that allograft fails more often did not separate these, and is dominated by processed tissue.
For non-irradiated tissue in adults, the trials and meta-analyses find equivalence. A 2013 review of eleven studies found no difference between autograft and non-irradiated, non-chemically-treated allograft in knee scores, laxity or failure. A 2014 systematic review of nine comparative studies found no difference in any outcome, with the explicit note that the findings applied to patients in their late twenties and early thirties. A 2016 meta-analysis of randomised trials found autograft better than allograft overall, but no difference on any of eight measures when the comparison was restricted to non-irradiated allograft. A 2018 meta-analysis of eleven trials and 1,172 patients reached the same conclusion: autograft better than irradiated allograft, no difference from non-irradiated. Individual randomised trials of fresh-frozen hamstring or tibialis allograft against hamstring autograft, in patients with mean ages around 30, found the same stability, the same knee scores and the same failure rates at two to eight years. A 2019 European consensus review put it plainly: similar low failure rates with non-irradiated, non-chemically-processed allograft and slower rehabilitation, but inferior results in young, highly active patients and with processed tissue.
For the young, high-demand athlete, the evidence goes the other way. A randomised trial in 99 mostly active-duty military patients, median age 26, compared fresh-frozen, non-irradiated tibialis allograft with hamstring autograft and followed them for at least ten years: 26.5% of allografts and 8.3% of autografts failed. In the MOON prospective cohort of 2,488 reconstructions, the odds of re-tear were 5.2 times higher with allograft than with patellar tendon autograft, and fell with each year of age. In the Kaiser registry, revision risk fell by about a third with every five years of age among allograft recipients. A Sydney series of 144 sterilised allografts, median age 26, had 5% failure at two years, every one in a patient aged 25 or under. And bone-block patellar tendon allografts, as distinct from soft-tissue allografts, carried 4.5 times the revision risk of autograft in the Kaiser registry at every age and with every processing method. A 2026 meta-analysis of soft-tissue grafts found no difference in revision or re-tear between allograft and autograft, with a small advantage to autograft on two knee questionnaires.
How that evidence is applied here. Non-irradiated, non-chemically-processed soft-tissue donor tendon is the mainstay in this practice for most patients, at any age: the recreational athlete returning to netball, football or skiing, the patient whose hamstrings measure small, the patient for whom kneeling pain from a patellar tendon graft would matter at work or in life, and revision and multi-ligament surgery where the patient’s own tendons are already spent. Professional and high-level competitive athletes typically have their own patellar tendon, often with a lateral tenodesis, because the cohorts in which allograft failed more were high-demand ones: active-duty military and competitive athletes with high activity scores. The line is drawn at the demand on the knee rather than at an age. Allograft rehabilitation is deliberately no faster than autograft rehabilitation, because the graft matures more slowly. The choice is explained before the operation, with these numbers, and recorded in your consent.
What does not depend on the graft
Arthritis in the long term: a meta-analysis of randomised trials found no difference in post-traumatic arthritis between graft types. Return to sport: the graft changes the details of rehabilitation, not the timeline, and the nine-month rule and the return-to-sport testing apply to all of them. And the single biggest determinant of whether the new ligament survives is not the tissue it was made from but what is done in the year afterwards, which is chapter 5.