Wisdom Teeth: Remove Them or Watch Them?

Reviewed 31 August 2026. Everything below is tied to a named source: three Cochrane reviews (one on removal versus retention of asymptomatic impacted wisdom teeth, one on surgical technique, one on post-operative pain relief), plus systematic reviews of antibiotics and of platelet-rich fibrin. Written for a patient deciding about a referral, and for clinicians. Not medical advice.

Related on this site: periodontitis and diabetes — because the distal of the second molar is a periodontal problem, not a “tooth problem”.

The short answers

• “Should my painless, disease-free impacted wisdom teeth come out?” The best existing evidence does not answer this confidently. The Cochrane review on exactly that question — removal versus retention of asymptomatic disease-free impacted third molars — found only two eligible studies (one randomised trial and one prospective cohort), reported no eligible study that measured health-related quality of life, and graded the evidence as low to very low certainty (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).
• What that one long-run cohort did show: in a subgroup of 416 men aged 24–84 followed from three years to more than 25 years, the presence of an asymptomatic disease-free impacted wisdom tooth may be associated with increased long-term risk of periodontitis on the distal of the neighbouring second molar — at very low certainty — while for caries risk on that tooth there was insufficient evidence of a difference (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).
• So the honest clinical frame is not “prophylactic extraction is unproven, therefore do nothing”. It is: extraction decisions are best driven by what is happening to the tooth in front of the wisdom tooth — probing depths, bleeding, radiographic bone loss, a distal defect, decay you cannot clean — and by symptoms. That is a monitoring question with a definite answer, not a philosophical one.
• Pain after surgery is manageable with ordinary analgesics, and the evidence is unusually specific: in the Cochrane review of seven trials with 2,241 participants, ibuprofen 400 mg beat paracetamol 1000 mg for at least 50% pain relief at six hours (RR 1.47, 95% CI 1.28 to 1.69, five trials) and reduced the need for rescue medication (RR 1.50, 95% CI 1.25 to 1.79, four trials); the paracetamol–ibuprofen fixed combination looked better still (RR 1.77, 95% CI 1.32 to 2.39, one trial, moderate quality) (Bailey et al., Cochrane Database Syst Rev 2013-12-01, PMID 24338830).
• On surgical technique, the Cochrane review of 62 trials with 4,643 participants is a warning about how thin the evidence is behind strong opinions: 33 studies (53%) at high risk of bias and 29 unclear, no usable data on the primary outcomes for coronectomy, and for flap design (envelope versus triangular) the intervals sat on both sides of no effect — alveolar osteitis OR 0.33 (95% CI 0.09 to 1.23, five studies, low certainty); wound infection OR 0.29 (0.04 to 2.06); permanent altered tongue sensation Peto OR 4.48 (0.07 to 286.49, one study) (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962).

Why this operation is controversial at all

Third molars are the teeth most often removed on the basis of a prediction rather than a present disease: the argument is that an impacted molar will eventually cause pericoronitis, resorb or destroy the second molar, decay, or form a cyst. The counter-argument is that the operation is not free: it carries short-term pain, swelling and trismus, and uncommon but serious risks — infection, dry socket (alveolar osteitis) and trigeminal nerve injury — with the risk of postoperative complications rising with age, as the Cochrane reviewers note when they point out that removal in older people carries more pain and complications (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796) (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962).

Add to that the health-system arithmetic: removing a tooth that never causes disease in a hundred people, to prevent a problem in a few, is a bet on risk — and a bet is only defensible when you can state the numerator. That is precisely what the “low to very low certainty” grading means here. It is not “the evidence shows no harm and no benefit”. It is: nobody has run the trial that would let a guideline make a firm recommendation either way, and the one long-run study available is at serious risk of bias (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).

The finding that actually changes behaviour: the second molar

The reason most oral surgeons watch rather than reassure is the distal surface of the second molar. An impacted third molar pressed against it creates a space that cannot be cleaned, a deep pocket, and often a resorption defect or a cavity that appears only once the damage is well advanced. In a retrospective study of 3211 consecutive extractions at a teaching hospital, patients re-examined 6 to 36 months after removal of an impacted mandibular third molar (158 subjects, mean age 29 ± 7 years) were found to have residual periodontal defects distal to the second molar, even though only 6% had a Community Periodontal Index score of 4 elsewhere in the mouth (Kan et al., J Clin Periodontol 2002-11-01, PMID 12472993). Read plainly: the damage is localised — a general “your gums are fine” is not evidence about that one site.

That gives the practical rule that replaces ideology. Ask what the numbers are at the distal of the second molar: probing depth, bleeding on probing, radiographic bone level, and whether there is a radiolucency on the distal root surface. If those are normal and stable over years, retention with monitoring is a defensible, evidence-consistent choice. If a pocket is deepening, bleeding, or bone is going, you now have a documented disease — and extraction, or at least a periodontal plan, is treatment rather than prophylaxis.

After the operation: what the evidence says about the boring parts

Pain control. A Cochrane review of the third-molar pain model included seven double-blind randomised trials, 2,241 participants; two at low risk of bias, three at high, two unclear. Ibuprofen was the superior analgesic at several doses: ibuprofen 400 mg versus paracetamol 1000 mg gave a risk ratio for ≥50% pain relief at six hours of 1.47 (95% CI 1.28 to 1.69) and for not needing rescue medication 1.50 (95% CI 1.25 to 1.79). The fixed combination of paracetamol 1000 mg with ibuprofen 400 mg showed RR 1.77 (95% CI 1.32 to 2.39) in one trial at moderate quality, with adverse events (nausea, vomiting, headache, dizziness) reported as comparable between groups, though the review could not formally analyse them (Bailey et al., Cochrane Database Syst Rev 2013-12-01, PMID 24338830). Practical translation: an anti-inflammatory-first plan is what the data support, combination therapy is better than either alone, and the combination needs your clinician’s judgement about stomach, kidney and bleeding risk — NSAIDs are not free for everyone.

Antibiotics. A systematic review and meta-analysis of randomised double-blind placebo-controlled trials — 21 studies pooled, 3,304 extractions — found systemic antibiotics significantly reduced the combined risk of dry socket and infection, overall RR 0.43 (95% CI 0.33 to 0.56) (Ramos et al., Oral Surg Oral Med Oral Pathol Oral Radiol 2016-10-01, PMID 27499028). Read it twice: the effect is real and sizeable on a complication rate; the decision whether to take that benefit routinely, given resistance and side-effect costs, is exactly the kind of thing the raw RR cannot decide for you. If you are being offered antibiotics, ask what the surgeon is trying to prevent in your case (duration of the operation, smoking, an infected space, a difficult impaction) — and if you are being refused them for a routine case, that too is a defensible reading of the same evidence.

Platelet-rich fibrin. Two reviews of the same idea, with an instructive difference in how much they could conclude. One evaluated PRF after mandibular third molar surgery: 1,430 publications screened, seven randomised trials for qualitative synthesis, only two poolable (485 extractions, 280 patients), with a reduction in alveolar osteitis (Canellas JVDS et al., Int J Oral Maxillofac Surg 2017-09-01, PMID 28473242); a second meta-analysis, in bilateral extractions, reported lower pain (SMD −0.53, 95% CI −1.02 to −0.05, with high heterogeneity I² = 75.7%), less swelling (WMD −0.55, 95% CI −1.08 to −0.01) and alveolar osteitis RR 0.35 (95% CI 0.16 to 0.75) (Xiang et al., BMC Oral Health 2019-07-01, PMID 31345203). That is a genuine signal on a nasty complication, from a small, heterogeneous base — which is why PRF is a reasonable offer, not a standard of care you should demand.

Technique. The Cochrane technique review’s headline is methodological: with 33 of 62 studies at high risk of bias, no usable primary-outcome data for coronectomy, and every flap, retractor, chisel-versus-handpiece and irrigation comparison returning “insufficient evidence” with intervals spanning no effect, there is no evidence-based answer to “which method is better” for most of the choices a surgeon makes (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962). What that means for you: the operator’s volume and familiarity with your specific impaction matter more than the brand of instrument, and questions about coronectomy, nerve risk and dry socket prevention are worth asking precisely because the literature cannot answer them for you.

What about nerve injury and “will I lose sensation?”

This is the risk that makes people hesitate, and it is a real one: the technique review lists permanent altered tongue sensation among the outcomes it could not resolve, with a single study giving a Peto OR of 4.48 with a confidence interval of 0.07 to 286.49 — a number that is formally meaningless and, honestly, informative about the state of the literature (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962). The clinically useful action is not to look for a pooled figure, it is to ask your surgeon to show you the radiograph and say what the relationship between the roots and the inferior alveolar canal looks like in your case (proximity, deflection of the canal, darkening, narrowing of the root — the signs that raise risk), and whether they would offer coronectomy instead. If a surgeon cannot discuss your film in those terms, get the second opinion you are already thinking about.

Evidence at a glance

How to decide, in a way that is defensible afterwards

• Symptomatic tooth — recurrent pericoronitis, a decay the tooth itself cannot be restored into, an unexplained cyst or lesion on the film: this is treatment of disease, and the evidence questions above do not really apply.
• Asymptomatic, disease-free, deep bony impaction in a 17-year-old: reasonable people differ; the evidence is low-certainty either way. Ask what your surgeon’s own threshold is, and how they will monitor the second molar if you keep the tooth.
• Distal pocket or bleeding at the second molar, or radiographic bone loss there: you now have a documented defect; the discussion is about a periodontal plan and whether removing the third molar improves access to it. Expect a discussion of probing depths, not of the wisdom tooth in isolation — see also what periodontal disease does to the rest of the body.
• Before orthodontics, before radiotherapy, before jaw surgery: these are the indications the Cochrane reviewers list as separate justifications, and they are decided by the treating team’s protocol rather than by the extraction-versus-retention literature (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).
• Age 40+: the same evidence base notes more pain and more complications with removal in older patients; if the tooth is healthy and silent, watchful monitoring with a defined interval tends to be the more defensible choice (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).

What the evidence does not support

• That routine removal of all asymptomatic impacted wisdom teeth prevents disease. There is no trial-grade support for that, and the Cochrane review’s certainty grading says so directly (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).
• That leaving them in is harmless either: there is a long-run signal about periodontal breakdown on the adjacent second molar, at very low certainty, and it is biologically coherent (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796) (Kan et al., J Clin Periodontol 2002-11-01, PMID 12472993).
• That “wisdom teeth cause crowding of the front teeth” is a reason to extract. Late lower incisor crowding appears among the justifications the review lists for prophylactic removal — and remains an unproven one in the same document (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).
• That any particular flap, suture, drain or irrigation protocol is superior: with 62 trials, the technique review could not establish it (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962).
• That antibiotics or PRF should be reflexively added — both reduce complications in pooled data; neither has an evidence base strong enough to make it a universal standard (Ramos et al., Oral Surg Oral Med Oral Pathol Oral Radiol 2016-10-01, PMID 27499028) (Xiang et al., BMC Oral Health 2019-07-01, PMID 31345203).

Frequently asked questions

My dentist says they must come out. Is that evidence-based? Ask which of two things they mean: there is disease (pericoronitis, decay, a defect on the second molar) — in which case it is treatment; or there may be disease later — in which case the honest answer is that the evidence for prophylactic removal is low to very low certainty (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796).

Is it true they can damage the tooth in front? That is the strongest and the least resolved concern: long-run association with periodontitis distal to the second molar (very low certainty) plus documented residual defects after extraction (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796) (Kan et al., J Clin Periodontol 2002-11-01, PMID 12472993).

How bad is the recovery? Expect a few days of pain, swelling and stiff jaw. The evidence supports ibuprofen as better than paracetamol, and the two together as better than either (Bailey et al., Cochrane Database Syst Rev 2013-12-01, PMID 24338830) — subject to your own stomach, kidney, blood-pressure and bleeding risks, which is a conversation with your prescriber.

Will I be numb? Permanent altered sensation is uncommon and not quantifiable from the reviewed evidence; the interval in the single study available spanned from essentially zero to enormous (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962). Ask to see your film and hear how close the roots are to the canal.

Should I take antibiotics afterwards? Pooled trials show they roughly halve the risk of dry socket and infection (RR 0.43), which is a real benefit against a real resistance and side-effect cost; ask your surgeon to explain the decision for your case (Ramos et al., Oral Surg Oral Med Oral Pathol Oral Radiol 2016-10-01, PMID 27499028).

Is PRF worth asking for? It reduced alveolar osteitis in pooled data (RR 0.35) with heterogeneity and small numbers — so it is reasonable to accept if offered, and reasonable to decline (Xiang et al., BMC Oral Health 2019-07-01, PMID 31345203).

What is coronectomy and should I want it? It is removing the crown and leaving the roots when they lie against the nerve. The technique review found no usable data for it on the outcomes that mattered (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962) — meaning it is a specialist judgement, and it should be discussed with an oral surgeon rather than chosen from a blog.

I am 45 and my teeth have never troubled me. That is close to the population where the balance shifts towards leaving them alone: the evidence base notes higher complication rates with age and provides no strong reason to intervene in an asymptomatic, disease-free tooth (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796). Set a recall and monitor the second molars.

Glossary

• impacted ↔ unable to erupt fully because of bone, tooth position or space; the direction and depth of the impaction predict the nerve question.
• pericoronitis ↔ inflammation of the gum flap over a partly erupted tooth; the commonest reason a “silent” tooth starts producing calls at 2am.
• alveolar osteitis (“dry socket”) ↔ painful loss of the clot in the socket, usually days 2–4; the outcome most pooled studies used.
• coronectomy ↔ deliberate retention of roots to protect the inferior alveolar nerve.
• prevented fraction / RR ↔ “how much lower the risk was”; RR 0.43 means roughly 43% of the events in untreated people remained.
• certainty of evidence ↔ how much the estimate can move (high/moderate/low/very low). For this whole topic: mostly low or very low. Read that as “the decision is yours and your clinician’s, not the literature’s”.

How this page was built, and what it cannot tell you

Sources were chosen because they are syntheses rather than single trials — the two Cochrane reviews that address the decision and the operation, the Cochrane analgesia review for aftercare, and two systematic reviews for the adjuncts that patients get asked to consent to. Bibliographic records were retrieved and verified programmatically from Europe PMC rather than recalled. Be aware of the shape of this literature: the review with the most clinical weight had two eligible studies (Ghaeminia et al., Cochrane Database Syst Rev 2020-05-01, PMID 32368796), and the one with the biggest participant count had unusable data for its own primary outcomes (Bailey et al., Cochrane Database Syst Rev 2020-07-01, PMID 32712962).

What it cannot tell you: whether your third molars will cause trouble, how close your roots are to a nerve, whether your second molar already has a defect, or how your own health conditions change the balance of NSAIDs, antibiotics and sedation. Every one of those is answered by a clinical examination plus a radiograph — and a written monitoring plan, if you keep the teeth.

This article summarises published research for information only. It is not medical or dental advice and does not replace assessment by a dentist or oral and maxillofacial surgeon. Seek urgent care for facial or neck swelling, fever, difficulty swallowing or breathing, or uncontrolled bleeding after surgery.

Sources

Peer-reviewed evidence

• Ghaeminia H, Nienhuijs ME, Toedtling V, Perry J, Tummers M, Hoppenreijs TJ, Van der Sanden WJ, Mettes TG. Surgical removal versus retention for the management of asymptomatic disease-free impacted wisdom teeth. Cochrane Database Syst Rev 2020-05-01;5():CD003879. doi:10.1002/14651858.cd003879.pub5 · PMID 32368796 · PMCID PMC7199383 · cited by 48 (Europe PMC)
• Bailey E, Worthington HV, van Wijk A, Yates JM, Coulthard P, Afzal Z. Ibuprofen and/or paracetamol (acetaminophen) for pain relief after surgical removal of lower wisdom teeth. Cochrane Database Syst Rev 2013-12-01;:CD004624. doi:10.1002/14651858.cd004624.pub2 · PMID 24338830 · PMCID PMC11561150 · cited by 54 (Europe PMC)
• Bailey E, Kashbour W, Shah N, Worthington HV, Renton TF, Coulthard P. Surgical techniques for the removal of mandibular wisdom teeth. Cochrane Database Syst Rev 2020-07-01;7():CD004345. doi:10.1002/14651858.cd004345.pub3 · PMID 32712962 · PMCID PMC7389870 · cited by 53 (Europe PMC)
• Ramos E, Santamaría J, Santamaría G, Barbier L, Arteagoitia I. Do systemic antibiotics prevent dry socket and infection after third molar extraction? A systematic review and meta-analysis. Oral Surg Oral Med Oral Pathol Oral Radiol 2016-10-01;122(4):403-425. doi:10.1016/j.oooo.2016.04.016 · PMID 27499028 · cited by 54 (Europe PMC)
• Xiang X, Shi P, Zhang P, Shen J, Kang J. Impact of platelet-rich fibrin on mandibular third molar surgery recovery: a systematic review and meta-analysis. BMC Oral Health 2019-07-01;19(1):163. doi:10.1186/s12903-019-0824-3 · PMID 31345203 · PMCID PMC6659259 · open access · cited by 43 (Europe PMC)
• Canellas JVDS, Ritto FG, Medeiros PJD. Evaluation of postoperative complications after mandibular third molar surgery with the use of platelet-rich fibrin: a systematic review and meta-analysis. Int J Oral Maxillofac Surg 2017-09-01;46(9):1138-1146. doi:10.1016/j.ijom.2017.04.006 · PMID 28473242 · cited by 50 (Europe PMC)
• Kan KW, Liu JK, Lo EC, Corbet EF, Corbet EF, Leung WK, Leung WK, Leung WK. Residual periodontal defects distal to the mandibular second molar 6-36 months after impacted third molar extraction. J Clin Periodontol 2002-11-01;29(11):1004-1011. doi:10.1034/j.1600-051x.2002.291105.x · PMID 12472993 · cited by 74 (Europe PMC)

Bibliographic records above were retrieved and verified programmatically from Europe PMC (authors, journal, volume, pages, DOI, PMID, citation count). Coverage, guideline and regulatory statements are quoted from the cited public documents with their dates; verify against the current version before relying on them. This article is not medical or dental advice.