Be in the know

Research on food safety is moving fast.

Homemade Vanilla Extract: Glass vs Plastic Storage and What the Research Says

Homemade vanilla extract is not a watery pantry syrup. It is a high-alcohol food, usually built on vodka or bourbon around 40 percent ethanol by volume, and that alcohol content is the real reason the glass vs plastic storage debate matters. Pure commercial extract sits at a legal floor of 35 percent ethanol. Regulators treat foods above 20 percent alcohol as a distinct packaging problem because ethanol pulls chemicals out of plastics more aggressively than water does. That is the demonstrated part of the story. What is not demonstrated is equally important: no peer-reviewed study was found that measured antimony, plasticizers, bisphenols, or other leachates from plastic bottles into vanilla extract itself. The concern is an informed extrapolation from ethanol content, food-contact testing rules, and closely related systems such as spirits in PET. A calm reading of that evidence still favors glass for long storage, and that is the case this article lays out.

What homemade vanilla extract actually is (and how much ethanol it holds)

Under the U.S. standard of identity, vanilla extract is the solution in aqueous ethyl alcohol of the extractable principles of vanilla beans. 21 CFR 169.175 requires ethyl alcohol of not less than 35 percent by volume, with at least one unit of vanilla constituent per gallon. There is no upper alcohol cap in that rule. The Alcohol and Tobacco Tax and Trade Bureau restates the same floor for formula work: a 1X vanilla extract is one unit per gallon with a minimum of 35 percent ethanol by volume (TTB, 2024).

Mass-market pure extracts usually sit at that statutory minimum. McCormick Pure Vanilla Extract lists alcohol at 35 percent on its ingredient statement (label transcription). Nielsen-Massey Mexican Pure Vanilla Extract lists water, alcohol (35 percent), sugar, and vanilla bean extractives (retail listing). Specialty bottles sometimes run higher; that is allowed, but it is brand-specific, not required.

Homemade practice almost always starts with an 80-proof spirit (40 percent ABV). Common kitchen recipes call for vodka or bourbon and note that commercial extract is typically 35 percent alcohol, or 70 proof (The Kitchn; Epicurious). If you want to make a batch yourself, our bean-and-bourbon method for homemade vanilla extract walks through beans, ratio, and cure time. Some people use 100-proof spirit (50 percent ABV). Beans add little water relative to the solvent, so the final ethanol content tracks the spirit. A practical homemade range is therefore about 35 to 50 percent ABV, most often around 40 percent. That places pure extract, store-bought or homemade, in the mid-to-high alcoholic food category, not with aqueous pantry items.

Two related products sit outside that frame. Bean-derived "vanilla flavoring" with ethanol below 35 percent by volume must be labeled under 21 CFR 169.177, not as extract. Imitation vanilla is usually water, propylene glycol, and vanillin, with little or no ethanol on commercial labels (Baker's listing; Serious Eats). The ethanol-driven packaging concern discussed below is mainly about pure extract, not about propylene-glycol imitation flavors.

Why ethanol matters for plastic food contact

Food-contact regulators do not treat "plastic in contact with food" as one problem. They grade extraction strength by food chemistry. Commission Regulation (EU) No 10/2011 assigns food simulants for that reason. Ethanol 10 percent (simulant A) stands in for hydrophilic foods. Ethanol 20 percent (simulant C) covers alcoholic foods up to 20 percent alcohol. Ethanol 50 percent (simulant D1) is assigned to alcoholic foods above 20 percent alcohol and to oil-in-water emulsions (Annex III). Industry guidance from PlasticsEurope walks through the same ladder: simulant A became 10 percent ethanol, C is 20 percent, and D1 is 50 percent (PlasticsEurope, 2011).

Pure vanilla extract at about 35 to 40 percent ethanol maps, under those rules, to the high-alcohol side of the table. Testing plastics intended for that kind of food uses 50 percent ethanol as the simulant. That is a regulatory assignment, not a claim that vanilla was poured into every lab bottle. Fifty percent is a conservative overestimate relative to typical extract, but it is the correct category, not the water category.

U.S. practice lines up. FDA chemistry guidance for food-contact substances generally uses 10 percent ethanol for aqueous, acidic, and low-alcoholic foods, and 50 percent ethanol for high-alcohol foods. The same guidance notes that migration into high-alcohol media is evaluated with 50 percent ethanol, and example data tables show higher migration in 50 percent ethanol than in 10 percent ethanol (FDA FCS Chemistry Recommendations). In short: both EU and U.S. systems treat higher ethanol as a stronger extractant of plastic-borne chemicals. That is demonstrated policy and chemistry guidance, not kitchen folklore.

What primary studies show in ethanolic systems (not in vanilla jars)

Again, the evidence gap first. No study in the research set measured leaching from plastic into vanilla extract specifically. The useful evidence is analogous: PET and other plastics in contact with ethanol solutions or spirits near the same alcohol strength as extract.

PET is the common clear plastic for beverage bottles. Antimony compounds are residual polymerization catalysts in PET. Westerhoff and colleagues documented antimony as a PET-related contaminant in bottled water systems (Westerhoff et al., 2008). Takahashi and colleagues likewise framed antimony in PET bottles as a catalyst residue with measurable release under certain conditions (Takahashi et al., 2008). That explains why antimony shows up in migration work. It does not, by itself, quantify risk in a vanilla jar.

Ethanol changes the polymer's behavior. Franz and Welle (2008) reported that higher-ethanol simulants, such as 50 percent and 95 percent ethanol, swell PET and speed migration of organic migrants relative to more aqueous conditions, and identified 50 percent ethanol as the appropriate worst-case simulant for PET beverage bottles (Food Additives & Contaminants, 2008). Filella's critical review of antimony in PET packaging summarizes the same theme: higher ethanol and longer contact time are among the factors that raise migration potential compared with water alone (Filella, 2020).

The closest beverage analog to homemade extract is spirits at roughly 40 percent ABV. Carneado and colleagues measured antimony migration into spirits and ethanolic media and found higher transfer than into water under comparable conditions (Carneado et al., 2017). A later review by the same group revisited antimony in PET-contact systems and again treated alcohol strength as a relevant driver of release (Carneado et al., 2023). Those papers support the analogy to extract. They do not measure extract.

For context on magnitude in a different matrix, Sánchez-Martínez and colleagues reported antimony migration from PET into EU food simulants in a range of roughly 0.5 to 1.3 micrograms per liter, all below the EU limit under the conditions they studied (Sánchez-Martínez et al., 2013). That is useful for scale and for avoiding alarm, not as a direct number for multi-year extract storage. Rungchang and colleagues modeled antimony migration from PET as a function of temperature and storage, reinforcing that time and heat matter in beverage systems (Rungchang et al., 2013). Heat is a separate stressor, and it is one reason Gadget Duke has written separately about what happens when PET water bottles get hot.

Soft plastics raise a different chemistry. In PVC food-contact work, phthalate plasticizers tend to migrate most into fatty media and into 50 percent ethanol simulants, more than into plain water (de Anda-Flores et al., 2021). That finding is about soft PVC and plasticizers, not about a typical PET spirit bottle. It still illustrates the broader packaging point: ethanol-rich foods are aggressive extractants for many organic migrants. For a material-by-material map of how food chemistry changes leaching behavior, see what chemicals can leach into food from common materials.

Glass vs plastic for storing vanilla extract: what the evidence supports

Commercial pure vanilla extract ships in glass for a reason that lines up with packaging science. Glass is chemically inert toward ethanol at room temperature under ordinary kitchen conditions. It does not swell the way PET can in higher-ethanol media, and it does not contribute residual polymerization catalysts such as antimony. That is why glass is the default industry package for high-alcohol extracts and for many spirits. On this site, glass also shows up as the inert baseline in other food-contact contexts, including glass meal-prep containers where heat and long contact are concerns.

Plastic is not a single material. PET beverage bottles, HDPE jugs, soft PVC tubing, and "BPA-free" polycarbonate replacements are different polymers with different residual catalysts and additives. "BPA-free" only means BPA was not used as the monomer of interest. It does not prove that a plastic is inert to 40 percent ethanol, nor that other migrants are absent. That distinction is the subject of a separate Journal piece on why BPA-free does not mean a container is automatically safe.

The fair, evidence-aligned takeaway for storing vanilla extract is conservative rather than dramatic. Pure extract is legally at least 35 percent ethanol and homemade batches are commonly about 40 percent. Regulators use 50 percent ethanol as the simulant for alcoholic foods above 20 percent ABV because higher ethanol extracts more from plastics. Primary studies show PET swelling in higher-ethanol simulants and higher migration of organics and, in some work, antimony into higher-ethanol media and spirits than into water. Direct measurements of vanilla stored for months or years in plastic were not found. Levels reported in some beverage studies sit below specific migration limits at ambient temperature, but extract is often stored for years, sometimes in reused bottles with high surface-to-volume ratios. Health risk magnitude for plastic-stored vanilla is therefore not quantified. What is quantified is the ethanol-driven mechanism and the analogy to spirits. Glass is the package that matches that evidence without requiring an untested leap.

Imitation vanilla in water and propylene glycol is a different composition problem. With little or no ethanol, the ethanol-swelling and high-alcohol simulant logic is weaker. That does not make every plastic ideal for imitation flavors. It does mean the specific high-alcohol concern that dominates pure extract storage is smaller for typical imitation formulas.

Practical framing without false certainty

A reasonable reading of the research is simple. If the goal is to minimize chemical transfer from the package into a multi-year, high-alcohol extract, glass is the package that industry and migration science already treat as inert for this class of food. Plastic bottles that work fine for water are not automatically equivalent for 35 to 40 percent ethanol. That statement does not require inventing a study of plastic vanilla bottles. It requires only the FDA alcohol floor, EU and FDA simulant rules, and the PET and plasticizer literature on ethanolic systems.

Claims that "plastic vanilla bottles leach toxins at known levels" would overshoot the evidence. Claims that "alcohol content is irrelevant to packaging" would undershoot it. The honest middle is the one that matches the sources: pure extract is a high-alcohol food; ethanol is a demonstrated migration driver; the vanilla-in-plastic measurement is missing; glass is the conservative, commercially standard choice supported by packaging science.

Sources