Reviewed 31 August 2026. The evidence base here is a Cochrane review of randomised trials, a systematic review and meta-analysis of longevity, and a peer-reviewed toxicology review of mercury vapour from amalgam. Written for patients deciding what to ask for, and for clinicians who want the effect sizes. Not medical advice.
The short answers
- For permanent back teeth, the best comparative evidence says amalgam fails less: in the Cochrane review, resin composites had a significantly higher risk of failure, risk ratio 1.89 (95% CI 1.52 to 2.35) — based mainly on two parallel-group trials covering 1,645 composite and 1,365 amalgam restorations, with all seven included trials judged at high risk of bias and the evidence graded low quality (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067).
- A separate meta-analysis covering trials and cohort studies with at least 12 months of follow-up (eight studies, published 1992–2013) reached the same direction: composite restorations in posterior teeth showed less longevity and more secondary caries than amalgam, with no statistically significant difference in fractures (Moraschini et al., J Dent 2015-09-01, PMID 26116767).
- The safety question, stated as the toxicology review states it: most standard amalgam formulations contain roughly 50% elemental mercury; mercury vapour is released and absorbed, and the number of amalgam surfaces correlates with urine mercury levels; but that review found “no convincing evidence pointing to any adverse health effects that are attributable to dental amalgam restorations besides hypersensitivity in some individuals” (Brownawell et al., Toxicol Rev 2005-01-01, PMID 16042501).
- So the honest framing is not “which material is safe” but a trade: a longer-lasting, cheaper-to-place, less technique-sensitive filling that is silver and requires more tooth removal to lock in — versus a tooth-coloured, bonded restoration that is easier to make leak-tight at the margins, more technique-sensitive, and more likely to need replacement. Cost, aesthetics, cavity location and your ability to keep the tooth dry at the appointment decide more than brand names do.
- Replacement of a sound, well-sealed amalgam “for safety” is not supported by any of these reviews. Removal of a failed or leaking restoration, on the other hand, is straightforward dentistry — and if you are pregnant, planning to remove intact amalgams is a conversation to have with your dentist rather than a decision to delegate to a blog.
What the two materials are, mechanically
Dental amalgam is a mixture: an alloy powder (silver, tin, copper, and historically zinc) triturated with elemental mercury, which then hardens by forming intermetallic phases. It contains no “liquid” mercury once set, but it does release small amounts of mercury vapour, mainly from the surface, with higher releases during chewing and, transiently, at placement and removal. Its clinical virtues are old and unglamorous: it is tolerant of moisture, it can be placed in a partially isolated field, it is strong in thin sections, and it wears at a rate broadly similar to tooth structure. Its problems: it needs a preparation with retention form (more tooth removed than a bonded restoration requires), it is visibly silver, it can grey the surrounding tooth structure, and it can cause galvanic currents against a neighbouring gold or copper crown.
Composite resin is a filler-loaded methacrylate that bonds to tooth structure via an etch-and-rinse or self-etch adhesive. The bond lets the preparation be smaller and lets the dentist rebuild a broken cusp rather than just fill a hole. The cost of that chemistry: the adhesive interface is the weakest link, the field must be isolated and dry, and technique sensitivity is not a slogan — it is the reason two identical-looking composites can have very different lifespans.
What the Cochrane review actually measured
The review set out to answer a narrow, useful question: for permanent posterior teeth, which material fails less? The authors searched five sources including the Cochrane Oral Health Group register, MEDLINE, EMBASE and LILACS with no language or date limits, and additionally contacted material manufacturers for unpublished studies. From 2,205 retrieved references they included seven trials (reported in ten articles), and excluded anything with less than three years of follow-up (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067).
The structure of the evidence matters as much as the result. Two of the trials were parallel-group studies contributing 1,645 composite and 1,365 amalgam restorations (921 children) to the analysis; the other five were split-mouth trials contributing 1,620 composite and 570 amalgam restorations, in an unclear number of children. Because of “major problems with the reporting of the data” in the split-mouth trials, the primary analysis rests on the two parallel-group trials, covering 3,265 composite and 1,935 amalgam restorations. All seven trials were judged at high risk of bias. The finding: resin restorations had a significantly higher risk of failure than amalgam, RR 1.89 (95% CI 1.52 to 2.35), and the authors’ own conclusion is framed as low-quality evidence that resin composites lead to higher failure.
Read that with two caveats in mind. First, “low-quality evidence” in Cochrane language means the true effect may be meaningfully different from 1.89 — it does not mean the comparison is worthless or that amalgam “wins”. Second, the population includes children with primary or newly erupted permanent teeth, where moisture control is hard and the benefit of a bonded material can be offset by technique failure; adult data with modern adhesives is a different animal, and the review’s own search window (to late 2013) predates a generation of bulk-fill and improved flowable materials.
The longevity meta-analysis, and where the difference shows up
A 2015 systematic review and meta-analysis widened the net beyond randomised trials: RCTs, controlled clinical trials and prospective and retrospective cohorts with at least 12 months of follow-up, comparing occlusal and occlusoproximal amalgam and composite restorations. From 938 PubMed/MEDLINE records, 89 CENTRAL titles and 172 Web of Science records, eight studies published between 1992 and 2013 were included, and all were rated high quality by the tool the authors used. The conclusions: composite restorations in posterior teeth still have less longevity and a higher number of secondary caries than amalgam, while for fractures there was no statistically significant difference between the materials across the follow-up times studied (Moraschini et al., J Dent 2015-09-01, PMID 26116767).
That “fractures, no difference” line deserves emphasis, because patient anxiety runs the other way: people fear the white filling will shatter the tooth. The evidence in these syntheses is that the composite itself is more likely to need re-intervention (debonding, marginal breakdown, recurrent decay at the margin), while catastrophic tooth fracture was not shown to differ between the two materials in the data pooled.
Mercury: what is measured, and what is not
The most useful neutral summary is a 2005 toxicology review written by a panel of toxicologists to assess the specific public concern. Its findings, in order: amalgam is a 150-year-old material and most standard formulations contain approximately 50% elemental mercury; experimental evidence “consistently demonstrates” that mercury vapour is released from amalgam and absorbed by the body; numerous studies report positive correlations between the number of amalgam restorations or surfaces and urinary mercury concentrations in non-occupationally exposed people; occupational studies show nervous-system and kidney effects at exposure levels that are substantially higher than those of people with fillings — although the review notes the exposure pattern differs (workers: 8 hours a day for 20–30 years; patients: 24 hours a day over part of a lifetime). Its conclusion: the review “uncovered no convincing evidence pointing to any adverse health effects that are attributable to dental amalgam restorations besides hypersensitivity in some individuals” (Brownawell et al., Toxicol Rev 2005-01-01, PMID 16042501).
Two honest limitations from the same paper. It explicitly identifies research gaps as the thing that would “definitively support or refute” the hypothesis — meaning the authors considered the question not closed in a scientific sense, merely unsupported by convincing evidence of harm at the exposures measured. And the immune question has its own older review: a 1995 survey of the literature on whether amalgam affects the immune system characterised the topic as controversial, with case reports of lichenoid reactions and allergy in sensitised individuals sitting alongside studies showing no systemic immune effect (Eneström et al., Int Arch Allergy Immunol 1995-03-01, PMID 7888781). If you have a documented metal allergy — particularly to mercury, or a lichenoid lesion adjacent to a restoration — that is a legitimate reason to choose a non-amalgam material, and it is a clinical finding, not a lifestyle theory.
On the environmental side, the Cochrane reviewers list disposal-related environmental impact as one of the two reasons interest in moving away from amalgam grew in the first place (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067). That is a policy and practice-management issue (amalgam separators, scrap handling) that affects your dentist’s equipment more than your mouth, but it explains why many offices stopped placing amalgam regardless of individual patient preference.
Which one should you choose
| Situation | What the evidence favours |
|---|---|
| Large posterior restoration where isolation is difficult (partially erupted molar, a child, a moist field) | Amalgam’s tolerance of technique problems is real, and the failure-rate difference in the Cochrane analysis is driven by exactly these settings (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067). |
| Visible area, or a patient who will object to a silver tooth | Composite. The longevity difference is a re-intervention risk, not an emergency, and aesthetics is a legitimate clinical criterion. |
| Tooth with a broken cusp that needs rebuilding | Adhesive composite (or an indirect restoration). Amalgam cannot bond a cusp back on. |
| Documented mercury or metal hypersensitivity, or a lichenoid reaction at a restoration margin | Non-amalgam material (Brownawell et al., Toxicol Rev 2005-01-01, PMID 16042501) (Eneström et al., Int Arch Allergy Immunol 1995-03-01, PMID 7888781). |
| Small occlusal cavity, adult, good isolation, cooperative patient | Either; ask the dentist which they place more often and how their own recall data looks. Material choice is the second decision; operator and isolation are the first. |
| “I want my old amalgams out for safety” | None of these reviews supports removing a sound restoration for mercury reasons, and removal itself transiently increases vapour release (Brownawell et al., Toxicol Rev 2005-01-01, PMID 16042501). Replace what has failed or leaked; discuss the rest. |
Cost, and how it actually shows up in a US estimate
Amalgam is usually cheaper to place: fewer minutes of chair time, no adhesive system, less need for perfect isolation. Composite costs more in the fee because it is time and materials, and — this is the part patients find unfair — it may also cost more over time in re-interventions, because that is precisely what the RR of 1.89 describes (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067). Where an employer or marketplace plan has a waiting period for “basic” restorative work, or a lower allowance for composite in posterior teeth (“amalgam-based fee allowance”), the difference can be tens to a couple of hundred dollars per tooth. If that matters to you, ask two concrete questions at the start: what is your plan’s allowance for a posterior composite, and is it coded by surfaces (D2140/D2330/D2391 for amalgam; D2330/D2391 for posterior composite, with one- to five-surface codes)? Then ask the dentist what they recommend on clinical grounds, and decide where to spend the difference.
What the evidence does not support
- That amalgam fillings cause chronic systemic disease. The toxicology review found no convincing evidence of adverse health effects attributable to amalgam other than hypersensitivity in some individuals — while also stating that research gaps remain before the hypothesis can be definitively refuted (Brownawell et al., Toxicol Rev 2005-01-01, PMID 16042501).
- That modern composites are equivalent to amalgam in every posterior situation. In the reviews above they were not, and the direction of difference (more re-intervention) was consistent.
- That removing intact amalgams improves health markers. That is a different literature; and removal has its own transient exposure and its own cost of new tooth structure lost.
- That “mercury-free” automatically means “safer restoration”: the material choice trades one set of risks for another (recurrent decay at the margin, technique sensitivity) (Moraschini et al., J Dent 2015-09-01, PMID 26116767).
- Any claim made for a specific brand of composite. These syntheses compare materials and settings, not marketing tiers.
Frequently asked questions
Are amalgam fillings safe? The reviewed evidence says mercury is released and absorbed, that urinary mercury correlates with the number of surfaces, and that no convincing adverse health effects have been shown besides hypersensitivity in some people (Brownawell et al., Toxicol Rev 2005-01-01, PMID 16042501). If you have metal allergy or a lichenoid reaction, that answer changes for you personally.
Do composite fillings last as long? In the pooled evidence for permanent back teeth, no: composites failed more, RR 1.89 (95% CI 1.52–2.35) in the Cochrane primary analysis, at low quality of evidence (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067); a separate meta-analysis found less longevity and more secondary caries with composite (Moraschini et al., J Dent 2015-09-01, PMID 26116767).
Do composites cause tooth fracture more often? The meta-analysis found no statistically significant difference in fractures between the materials over the follow-up studied (Moraschini et al., J Dent 2015-09-01, PMID 26116767).
Is it true that amalgam needs more drilling? Yes — a conventional amalgam preparation requires retention form; composite can be placed with a more conservative preparation because it bonds.
Should I have old amalgams removed? If the restoration is intact and asymptomatic, none of these reviews supports removal for safety reasons; removal also transiently raises vapour release (Brownawell et al., Toxicol Rev 2005-01-01, PMID 16042501). If it is cracked, leaking, or has decay at the margin, replace it — and then you can choose the material.
Why can’t I get amalgam anymore? Some offices dropped it for environmental-compliance and demand reasons, some for aesthetic demand. Ask directly: if your case benefits from it and your plan pays less for composite in a molar, you are entitled to a practice that places both.
Will insurance pay less for composite? Commonly yes in back teeth: plans often allow an amalgam-based fee when composite is chosen for aesthetics, and you pay the difference. Ask for the allowance before the appointment.
Children — does it matter? In the Cochrane data, a large share of the analysed restorations came from children (921 in the two parallel trials), and this is the population where technique sensitivity and moisture control most erode composite’s advantage (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067). Stainless steel crowns or sealants may be the better answer for a badly broken primary molar anyway.
Is grey staining of the tooth a problem? It is a cosmetic effect of amalgam on surrounding structure, not a sign of disease; it is one of the reasons patients switch to composite at the next re-intervention.
What about the environment? The Cochrane review lists environmental impact of disposal among the reasons amalgam use came under scrutiny (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067). That is handled by separators and scrap handling in the office, and is a legitimate reason to prefer an office that invests in them.
Glossary: chair words ↔ billing words ↔ plain words
- “the filling fell out” ↔ restoration failure: an umbrella covering fracture, debonding, loss of retention, recurrent decay and wear. In the Cochrane review failure was the primary outcome and RR 1.89 refers to that composite of reasons, not to one mechanism (Rasines Alcaraz et al., Cochrane Database Syst Rev 2014-03-01, PMID 24683067).
- secondary caries / recurrent decay ↔ new decay at the margin of an existing restoration — the difference the 2015 meta-analysis found favouring amalgam (Moraschini et al., J Dent 2015-09-01, PMID 26116767).
- occlusal (one-surface) vs MO/MOD (two/three-surface) ↔ the surface count that determines the procedure code, and usually the price.
- isolation / rubber dam ↔ keeping the field dry; the single most important variable the composite side of the trade depends on.
- RR and 95% CI ↔ risk ratio with an interval: 1.89 (1.52–2.35) means the composite group’s failure risk was somewhere between 1.5 and 2.4 times that of amalgam, in the studies analysed.
- low-quality evidence ↔ Cochrane’s certainty grade meaning the estimate is unreliable, not that it is worthless or reversed.
How this page was built, and what it cannot tell you
Sources were chosen for directness of comparison: the Cochrane review for randomised evidence on failure, the 2015 meta-analysis for longevity and secondary caries across a wider set of study designs, and the 2005 toxicology review for the mercury question because it is the kind of document that argues both sides (documented release and correlation with urinary mercury; no convincing evidence of adverse effects except hypersensitivity; research gaps named). Bibliographic details for each item below were retrieved and verified programmatically from Europe PMC rather than written from memory.
What it cannot tell you: whether your specific cavity can be isolated well enough for a bonded restoration (that is a chairside judgement), how long your own filling will last given your hygiene, sugar frequency and bruxism, whether the amalgam in tooth 30 is leaking, or what your plan will allow. It can tell you that the choice is a trade of durability against conservative preparation and appearance, that the material debate is not a safety debate in the reviewed literature — and that a dentist who cannot discuss the trade in those terms is making a decision for you instead of with you.
This article summarises published research for information only. It is not medical or dental advice and does not replace an examination by a licensed dentist. Do not have restorations removed or placed on the strength of a web page, including this one; seek prompt care for pain on biting, swelling, or a broken tooth.
Sources
Peer-reviewed evidence
- Rasines Alcaraz MG, Veitz-Keenan A, Sahrmann P, Schmidlin PR, Davis D, Iheozor-Ejiofor Z. Direct composite resin fillings versus amalgam fillings for permanent or adult posterior teeth. Cochrane Database Syst Rev 2014-03-01;:CD005620. doi:10.1002/14651858.cd005620.pub2 · PMID 24683067 · cited by 79 (Europe PMC)
- Moraschini V, Fai CK, Alto RM, Dos Santos GO. Amalgam and resin composite longevity of posterior restorations: A systematic review and meta-analysis. J Dent 2015-09-01;43(9):1043-1050. doi:10.1016/j.jdent.2015.06.005 · PMID 26116767 · cited by 132 (Europe PMC)
- Brownawell AM, Berent S, Brent RL, Bruckner JV, Doull J, Gershwin EM, Hood RD, Matanoski GM, Rubin R, Weiss B, Karol MH. The potential adverse health effects of dental amalgam. Toxicol Rev 2005-01-01;24(1):1-10. doi:10.2165/00139709-200524010-00001 · PMID 16042501 · cited by 65 (Europe PMC)
- Eneström S, Hultman P. Does amalgam affect the immune system? A controversial issue. Int Arch Allergy Immunol 1995-03-01;106(3):180-203. doi:10.1159/000236843 · PMID 7888781 · cited by 65 (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.