The Ames test is a rapid laboratory assay that uses specially engineered bacteria to determine whether a chemical can cause mutations in DNA. Published by Bruce Ames in 1973, it became one of the most widely adopted safety screening tools in toxicology because it was cheap, fast, and remarkably good at flagging chemicals that might cause cancer. Regulatory agencies around the world now require it as a first-line screen for new pharmaceuticals, food additives, pesticides, and industrial chemicals, and it remains the backbone of genetic toxicology nearly five decades later.
How the Test Works
The core idea is deceptively simple. The bacteria used in the Ames test have been deliberately crippled: they carry mutations that prevent them from making an amino acid they need to survive, usually histidine. When you spread these bacteria on a plate that contains only a trace of histidine, they run out quickly and stop growing. But if a chemical on the plate damages their DNA in exactly the right way, it can reverse the original mutation and restore the bacteria’s ability to make histidine on their own. Those rare “reverted” bacteria keep dividing and form visible colonies that you can count. More colonies means the chemical is causing more mutations.
The strains of Salmonella typhimurium used in the standard test each carry a different mutation in genes of the histidine operon, and these mutations serve as hot spots for mutagens that damage DNA through different mechanisms. Some strains detect chemicals that swap one DNA base for another (base-pair substitutions), while others catch chemicals that insert or delete bases (frameshift mutations). An Escherichia coli strain is also used; it carries a mutation in the tryptophan gene and picks up mutagens that target a different type of DNA site.
1PubMed. Bacterial mutagenicity assays: test methodsThe S9 Metabolic Activation Step
Many chemicals are not dangerous in the form you swallow or inhale them. They become harmful only after your liver enzymes convert them into reactive by-products. Bacteria lack these enzymes, so testing a chemical on bacteria alone would miss an entire class of hazards. To solve this, the Ames test adds a liver extract, called the S9 fraction, to the plate. This extract is typically prepared from rat liver and contains the same enzyme family (cytochrome P450) that processes drugs and toxins in mammals. The S9 mix converts “pro-mutagens” into their active forms right there on the plate, so the bacteria are exposed to the same metabolites a mammalian body would produce.
2PubMed. Advantage of the use of human liver S9 in the Ames testEvery Ames test is run in parallel with and without S9. A chemical that only triggers colonies on the S9-containing plates is flagged as a pro-mutagen, one that needs metabolic activation. This distinction matters for regulatory follow-up because it tells toxicologists something about the chemical’s mechanism. Compounds like polycyclic aromatic hydrocarbons, aromatic amines, and nitrosamines are classic examples of pro-mutagens that would be missed entirely without S9.
One limitation of using rat liver S9 is that human liver enzymes do not always activate the same chemicals or produce the same metabolites. Research has explored whether human liver S9 might give results more relevant to human risk, and for certain compound classes the answer is yes, though rat S9 remains the regulatory default because it has been validated over decades.
3PubMed. Advantage of the use of human liver S9 in the Ames testWhich Bacterial Strains Matter Most
The international test guideline (OECD Test Guideline 471) calls for at least five strains of bacteria, covering both base-pair substitution and frameshift mutations. In practice, two Salmonella strains, TA98 and TA100, do most of the heavy lifting. An analysis carried out under the International Workshop on Genotoxicity Testing found that TA98 and TA100 together, with and without S9, detected about 93% of all the mutagens identified by the full panel of recommended strains.
4PubMed Central. Analysis of chemical structures and mutations detected by Salmonella TA98 and TA100The remaining strains exist to catch the mutagens that TA98 and TA100 miss. Some chemicals target AT base pairs specifically, and for those you need a strain that has an AT site at its critical mutation point. This is where the E. coli WP2 strains come in. They carry a tryptophan mutation with an AT base pair at the reversion site and are used either alongside or as a substitute for the Salmonella strain TA102, which also has an AT target site.
5PubMed Central. Analysis of chemical structures and mutations detected by Salmonella TA98 and TA100For specific compound classes, researchers sometimes fine-tune the strain selection. Work on nitrosamine impurities in pharmaceuticals, for instance, showed that strains TA100 and TA1535 combined with E. coli WP2uvrA(pKM101), hamster-derived S9, and a 30-minute pre-incubation step gave the most sensitive detection of common nitrosamines like NDMA and NDEA.
6PubMed Central. Ames test study designs for nitrosamine mutagenicity testing: qualitative and quantitative analysis of key assay parametersHow Well Does It Predict Cancer?
The Ames test became famous largely because of its link to carcinogenicity. The logic goes like this: most known chemical carcinogens work by damaging DNA, and the Ames test detects DNA damage. So a positive Ames result raises a red flag that the chemical might cause cancer in animals or people. This relationship is grounded in what is called the somatic mutation theory of carcinogenesis.
7PubMed. A perspective on the development of the Ames Salmonella/mammalian-microsome mutagenicity assayFor chemicals that directly damage DNA, the predictive value is genuinely impressive. A large analysis of nitrosamine compounds found that the Ames test’s sensitivity for predicting rodent carcinogenicity ranged from 93 to 97%, meaning it caught nearly all the carcinogens in the dataset. The specificity ranged from 55 to 86%, meaning it also flagged some chemicals that turned out not to be carcinogenic in rodents.
8PubMed. Use of the bacterial reverse mutation assay to predict carcinogenicity of N-nitrosaminesThat high sensitivity holds for DNA-reactive chemicals broadly. But the relationship weakens sharply when you look at chemicals that cause cancer through other routes, such as chronic inflammation, hormonal disruption, or immune suppression. These “non-genotoxic carcinogens” do not directly damage DNA and therefore tend to come up negative in the Ames test. A review of carcinogenicity prediction concluded that no reliable mutagenicity-based alternative exists for assessing the risk of non-DNA-reactive chemicals, and that the Ames test is useful primarily for the subset of carcinogens that work by mutating genes.
9PubMed. Alternatives to the carcinogenicity bioassay: in silico methods, and the in vitro and in vivo mutagenicity assaysAnd there is another wrinkle: mutagenic potency in the Ames test does not reliably predict how potent a carcinogen will be in animals. A chemical can light up a plate with thousands of revertant colonies but turn out to be a weak carcinogen, or vice versa. Research attempting to use the quantitative strength of Ames results to rank carcinogenic risk found the relationship to be, at best, weak.
10Environmental and Molecular Mutagenesis. Predicting rodent carcinogenicity from mutagenic potency measured in the Ames Salmonella assayWhere the Ames Test Falls Short
The test is designed around bacteria, and bacteria are not people. They lack the complex DNA repair machinery, cell cycle checkpoints, and tissue organization of mammalian cells. A chemical might damage bacterial DNA but be efficiently repaired in a human cell before it ever causes harm. Conversely, some chemicals that are harmless to bacteria can still damage mammalian chromosomes in ways the Ames test cannot detect, like breaking chromosomes or interfering with the mitotic spindle.
This is why regulatory agencies never use the Ames test alone. It typically sits as the first test in a battery that includes at least one mammalian cell assay, such as a micronucleus test or a chromosome aberration test, to catch the types of genetic damage bacteria cannot report on.
11PubMed. Genotoxicity assessment of triclocarban by comet and micronucleus assays and Ames testFalse negatives are a documented concern. Many chemicals that test negative in the Ames test still induce tumors in rats or mice. An analysis of the National Toxicology Program’s two-year rodent cancer study database noted that a number of Ames-negative chemicals nonetheless caused tumors, particularly lung tumors in mice exposed by inhalation, a pattern linked to tissue-specific effects that a bacterial assay simply cannot model.
12Toxicology Research and Application. The “false-positive” conundrum in the NTP 2-year rodent cancer study databaseNanomaterials present another challenge. Many nanoparticles are too large to cross bacterial cell walls, so even if they would damage DNA inside a mammalian cell, they may produce a negative Ames result simply because they never reached the bacteria’s DNA. Researchers have recommended modifications to the standard protocol to improve nanoparticle uptake, but these are not yet part of the official guideline.
13PubMed Central. In vitro genotoxicity testing strategy for nanomaterials and the adaptation of current OECD guidelinesRegulatory Standing
The OECD Test Guideline 471 codifies the Ames test as the “bacterial reverse mutation test” and lays out its requirements: at least five strains, testing with and without S9, at least five concentrations of the test chemical, and appropriate positive and negative controls. This guideline is accepted by regulatory bodies across the world, making the Ames test one of the most internationally harmonized safety assays in existence.
14OECD Publishing. Test No. 471: Bacterial Reverse Mutation TestIn pharmaceutical development, a positive Ames result is effectively a stop sign. Regulatory agencies treat bacterial mutagenicity as strong evidence that a compound damages DNA, and drug candidates that trigger a positive result face intensive follow-up testing or outright rejection. This is part of why the Ames test occupies such an outsized role in safety screening: it is the cheapest and fastest way to flag a potential deal-breaker early in the development pipeline.
15PubMed. The test that changed the world: The Ames test and the regulation of chemicalsIn practice the Ames test is usually the first step in a three-test genotoxicity battery. A recent review of data from this battery for pesticide-related chemicals argued that the mammalian cell gene mutation assay (the third test in the battery) was not adding meaningful safety signal beyond what the Ames test and the micronucleus assay already provided, and was generating unnecessary follow-up animal studies.
16PubMed Central. Assessment of the three-test genetic toxicology battery for groundwater metabolitesEnvironmental Monitoring
Beyond pharmaceutical safety, the Ames test has found a second life as a tool for environmental monitoring. Researchers use it to screen soil extracts, drinking water concentrates, and industrial effluents for the presence of mutagenic contaminants. This is particularly useful when you need a yes-or-no answer to the question “is this environmental sample hazardous?” without identifying every individual chemical in a complex mixture.
Studies of contaminated soil in industrial regions have paired the Ames test with chemical analysis of polycyclic aromatic hydrocarbons to determine whether soil contamination poses a genetic hazard.
17PubMed. Soil contamination detected using bacterial and plant mutagenicity tests and chemical analyses Similar work on drinking water sources has shown that organic extracts from raw water can induce DNA damage in the Ames test, sometimes even without S9 activation, signaling that the mutagens in the water are already in their active form.
18PubMed Central. Evaluation of Genotoxic and Mutagenic Activity of Organic Extracts from Drinking Water SourcesSoil leachates from sites contaminated with PAHs and heavy metals have also tested positive in the Ames test, raising concerns about threats to the surrounding aquatic environment as contaminants wash into groundwater and streams.
19Organic Geochemistry. Use of in vitro (Ames and Mutatox tests) and in vivo (Amphibian Micronucleus test) assays to assess the genotoxicity of leachates from a contaminated soilMiniaturized and High-Throughput Versions
The traditional Ames test uses Petri dishes, requires substantial quantities of the test chemical, and has throughput limits that make it impractical for screening large libraries of compounds. Over the past two decades, miniaturized versions have emerged that run the same biology in 96-well or 384-well microplates filled with liquid medium instead of agar. The readout switches from counting colonies to a simple color change, making it amenable to automated plate readers and even robotic pipetting systems.
20PubMed. The Ames II and Ames MPF™ Tests: Miniaturized Liquid Microplate Formats of the Ames Test in Petri DishesThese miniaturized formats serve several purposes. They use less test chemical, which matters when you are screening an early-stage drug candidate and only have milligrams of material. They use less S9, reducing the number of animals needed to produce liver extract and aligning with the 3Rs principles of replacement, reduction, and refinement in animal research. And they enable high-throughput screening during early drug discovery, so genotoxic compounds can be weeded out before significant resources are invested.
21PubMed. Investigation of chemicals with inconsistent Ames results using miniaturized Ames test systemsAn interesting practical question is whether these miniaturized formats agree with the traditional plate test. For the large majority of chemicals, they do. But some compounds give inconsistent results between formats, and these edge cases are the subject of ongoing investigation to determine whether the discrepancies reflect genuine biological variability or artifacts of the miniaturized protocol.
22PubMed. Investigation of chemicals with inconsistent Ames results using miniaturized Ames test systemsComputational Models Trained on Ames Data
The Ames test has generated one of the largest curated datasets in toxicology, with results for thousands of chemicals publicly available. This has made it a favorite training ground for computational prediction models. The idea is appealing: if you can predict from a chemical’s structure alone whether it will be Ames-positive, you could skip the bench experiment entirely for early screening.
Two main approaches dominate. One uses quantitative structure-activity relationship (QSAR) models, which learn statistical patterns between molecular features and Ames outcomes. The other uses rule-based expert systems, which encode known structural alerts, essentially molecular red flags associated with mutagenicity, and check new chemicals against that knowledge base. The accuracy of these in silico models has been reported to approach about 85%, which is roughly the same as the Ames test’s own reproducibility when the same chemical is tested in different laboratories.
23Oxford Academic. Refinement, Reduction, and Replacement of Animal Toxicity Tests by Computational MethodsMore recent work has pushed into deep learning. Some models are trained not just on the overall positive/negative Ames outcome but on strain-specific results, learning which structural features trip up TA98 versus TA100, for example. Others use advanced architectures to handle “problematic” molecular species like aromatic amines that require metabolic activation, where simpler models have historically struggled.
24PubMed. Multiple Instance Learning Improves Ames Mutagenicity Prediction for Problematic Molecular SpeciesRegulators are cautiously embracing these tools. Computational predictions are increasingly accepted as supporting evidence in safety dossiers, particularly for impurities present at very low levels where running a full Ames test may not be practical. But no agency currently allows an in silico prediction to fully replace an actual Ames test for a new drug candidate. The computational models are seen as a complement, not a substitute, useful for prioritizing which compounds to test and for filling gaps when physical testing is not feasible.
The Hair Dye Episode and Consumer Safety
One of the earliest and most publicly visible applications of the Ames test came not from a regulatory mandate but from Bruce Ames’s own laboratory. In 1975, Ames published a study showing that commercial hair dye formulations were mutagenic. The finding attracted widespread media attention and is credited with catalyzing the NIEHS testing program for environmental chemicals.
25PubMed. Reflections on a career and on the history of genetic toxicity testing in the National Toxicology ProgramThe hair dye episode illustrates both the power and the limits of the test. The Ames result demonstrated that the dye chemicals could damage DNA in bacteria, which was alarming. But translating that finding into actual cancer risk for people who color their hair proved far more complicated. Decades of epidemiological studies on hairdressers and consumers have produced mixed results, with some studies finding modest increases in certain cancers and others finding none. The chemicals themselves have been reformulated multiple times since the 1970s, partly in response to mutagenicity data. The episode drove real change in consumer product safety, even though the leap from a bacterial plate to human cancer was never straightforward.
Measuring Mutagenic Potency
A positive or negative Ames result is useful, but researchers and regulators sometimes want more: how mutagenic is the chemical, quantitatively? The standard approach is to plot the number of revertant colonies against the dose and calculate a slope, typically expressed as revertants per nanomole of compound. The steeper the slope, the more potent the mutagen.
The range of potencies across known mutagens is enormous, spanning more than seven orders of magnitude in the Ames reversion test.
26Mutation Research/Reviews in Genetic Toxicology. Genotoxic activity and potency of 135 compounds in the Ames reversion test and in a bacterial DNA-repair test This wide spread means that a simple “positive” label groups together chemicals with wildly different abilities to damage DNA, which limits the usefulness of a binary result for risk assessment.
Complicating things further, both spontaneous and chemically induced mutation rates vary between laboratories and even within the same lab over time. Researchers have explored whether incorporating the background mutation rate into the potency calculation yields more consistent numbers, and the evidence suggests it helps, though no single potency metric has become universally standard.
27Mutation Research. A comparison of alternative measures of mutagenic potency in the Salmonella (Ames) test