Every fact on this site, with its evidence grade, its caveats, and a link to the source.
We start you on the strong ones: findings that have been replicated or come from one large, well-controlled study. Moderate and weak facts are here too, labeled for what they are. Use “Link to this fact” to share a single card.
Indoor air / Measured in people or real products
The UK home with the heaviest, most frequent use of cleaning and fragranced products (9 different products, some used more than 10 times in a week, plus a scented candle) had a 5-day average D-limonene concentration of up to 1,439 ug/m3, around 80 times the study's median home and possibly the highest domestic value reported in the literature at the time.
Moderate evidence: Single home out of 25 driving the extreme; 5-day average, not a short-term peak, which the authors state was likely higher.
Source: Wang 2017, Environmental Science: Processes & Impacts Link to this fact
Indoor air / Measured in people or real products
A flickering (flame-disturbed) candle can emit up to 1,700 times more soot than the same candle burning steadily and undisturbed, based on chamber measurements of one candle type under both conditions.
Moderate evidence.
Caveats
Comparison is for a single candle type (candle 1) with one steady-burn control experiment versus two replicate stressed-burn experiments; the 60-1700x range reflects the size of the effect across the measurement, not a single precise multiplier.
Indoor air / Measured in people or real products
Candle wax and wick choice makes up to a 27-fold difference in soot, fine-particle, and particle-bound PAH emissions between candle types, but essentially no difference in NOx, formaldehyde, or gas-phase PAH emissions, in a controlled chamber study of five unscented pillar candles.
Moderate evidence.
Caveats
Five candle types, two replicate experiments each, single chamber, and single research group; not verified across other candle shapes (e.g. tapered, container candles) or other labs.
Indoor air / Measured in people or real products
In unscented candles, formaldehyde and other carbonyl emissions stayed flat and low (39-54 micrograms per hour) regardless of wax type, wick type, or whether the flame was steady or flickering, which the study's authors attribute to the absence of added fragrance or dye, since those additives (not the wax or wick) are understood to be the main source of a candle's aldehyde emissions.
Moderate evidence.
Caveats
The attribution to fragrance/dye is the authors' interpretation, citing prior literature (Derudi et al. 2014), not a within-study comparison against scented versions of the same candles; this study tested no scented candles.
Indoor air / Measured in people or real products
Burning a scented candle releases far more organic combustion by-products than burning an unscented candle of the same wax: total VOC emission rates reached up to 14,388 micrograms per candle per hour for scented candles versus 70-204 for the four unscented candles tested (24 fuel/fragrance combinations, chamber study).
Moderate evidence.
Caveats
Single lab (Fraunhofer WKI), industry/candle-association and fragrance-house funded (European Candle Association, National Candle Association, Latin American Candle Manufacturers Association; advisory committee incl. Arylessence, Belmay, Firmenich, Givaudan, IFF, Symrise, Takasago); n=4 candles per combination, purpose-built research candles not verified as representative of specific retail products.
Source: Salthammer 2021, Environment International Industry-funded Link to this fact
Indoor air / Measured in people or real products
Benzo[a]pyrene, a combustion-linked carcinogen, was measurable in 5 of 24 candle-burning experiments (up to 12 nanograms per candle per hour) and only in scented candles; none of the 4 unscented candles produced a detectable amount.
Moderate evidence: Small n (5/24 detections); detection near the analytical limit of quantification, which the authors themselves note adds uncertainty to the calculated exposure comparisons.
Source: Salthammer 2021, Environment International Industry-funded Link to this fact
Indoor air / Measured in people or real products
The fragrance added to a candle, not the wax it is made from, is the main driver of extra combustion emissions: unscented candles made from four different waxes (palm, paraffin, soy, stearin) formed the lowest-emission cluster in the study's statistical analysis, while scented versions of the same waxes released far more VOCs, PAHs, and formaldehyde.
Moderate evidence.
Caveats
Based on a hierarchical cluster analysis of 24 experiments in one chamber study; individual scented fragrance families varied in how much they raised emissions (floral, oriental, and spice & edibles raised emissions less than fresh and fruit).
Source: Salthammer 2021, Environment International Industry-funded Link to this fact
Indoor air / Measured in people or real products
The estimated respiratory tract deposited dose of these new nanoparticles from scented wax melt use (median 2.9 x 10^10 particles per minute) was comparable to the dose rates reported for burning a scented candle, running a gas stove, or diesel and natural gas engine exhaust, and about 1,000 times higher than the same test house's background air.
Moderate evidence.
Caveats
Modeled dose rate from measured particle size distributions using a published respiratory deposition method, not a direct measurement of dose in a human airway; combustion-source comparison values come from separate studies with different instruments and settings.
Source: Patra et al. 2025, Environmental Science & Technology Letters Link to this fact
Indoor air / Measured in people or real products
In a lab chamber test, three of six candles — a strawberry candle, a clean-cotton candle, and a plain (unscented) reference candle — produced formaldehyde levels above both the ACGIH occupational guideline (325 ppb) and OSHA's legal limit (730 ppb) while lit, up to 2,098 ppb for the strawberry candle.
Moderate evidence.
Caveats
These are concentrations from a small impinger/chamber system fed by 2 L/min of pure air pulled directly past the flame, not measured or modeled room air; single lab, six specific retail products, no replication.
Source: Ahn et al. 2015, Journal of Hazardous Materials Link to this fact
Indoor air / Measured in people or real products
One kiwi-melon scented candle emitted more total measured VOC while sitting unlit (12,742 ppbC) than while it was actually burning (2,766 ppbC) — more than 4 times as much — and the unlit figure was comparable to a published emission rate for a carpet cleaner.
Moderate evidence.
Caveats
TVOC is a summed, toluene-equivalent proxy, not a hazard-weighted measure; the carpet-cleaner comparison is to a different study's figures for a different product, not a side-by-side measurement.
Source: Ahn et al. 2015, Journal of Hazardous Materials Link to this fact
Indoor air / Measured in people or real products
Using a mass-balance model of a realistic room scenario (one person present 8 h, two candles lit for 4 h, 20 cubic-meter room, 0.5 air changes per hour), Kim et al. 2016 calculate an 8-hour time-weighted-average formaldehyde concentration exceeding 100 ppb in the worst case, more than 6 times NIOSH's 16 ppb 8-hour recommended exposure limit.
Moderate evidence.
Caveats
This is a mass-balance model calculation built on chamber-derived emission factors, not a direct room measurement; the two-candle, 4-hour worst-case assumptions may not reflect typical single-candle household use.
Source: Kim et al. 2016, Journal of Hazardous Materials Link to this fact
Indoor air / Animal study
In a mouse chamber study modeling occupational candle exposure, burning scented candles for 4.5 hours a day, 5 days a week, produced measured chamber levels of 2157.5 ng/m3 PAHs, 181 micrograms per cubic meter VOCs, and 219,000 nanoparticles per cubic centimeter (median size 10.7 nanometers) during active burning.
Moderate evidence.
Caveats
This is an occupational-intensity exposure chamber (hours per day, days per week, for 8 weeks), not a measurement of typical home candle-burning exposure; single lab, single candle type (brand/composition undisclosed).
Source: Chandrasekaran 2021, Current Research in Toxicology Link to this fact
Indoor air / Animal study
An "unscented" candle was not toxicologically inert: its headspace was 51.65% toluene by GC-MS peak area, and burning it in rats for 8 weeks still caused duration-dependent serum inflammation and lung injury compared with fresh air, though consistently less severe than the scented candle at every exposure duration tested.
Moderate evidence: One unscented candle product; wax/wick composition not disclosed; toluene here is an emission from the candle itself, not an added fragrance ingredient.
Source: Mohamed et al. 2025, Frontiers in Public Health Link to this fact
Indoor air / Other evidence
Burning a scented candle at low ventilation (0.5 air changes per hour) produced significantly higher total PAH and cancer-risk-weighted PAH emissions than at 1.0 or 2.0 air changes per hour, in a controlled chamber study.
Moderate evidence.
Caveats
Single lab, single scented-candle formulation (paraffin + 6% lavender oil), small chamber; total PAH itself was not monotonic with ACH (0.5 > 2.0 > 1.0), only the toxicity-weighted BaPeq fraction fell monotonically.
Indoor air / Other evidence
Relative humidity had a bell-shaped, non-monotonic effect on a scented candle's PAH emissions, peaking at 75% RH rather than at the lowest or highest humidity tested.
Moderate evidence.
Caveats
Single lab, single candle formulation; mechanism (flame-temperature/water-vapor effects) is inferred from indirect combustion indicators, not directly measured — authors call it a plausible hypothesis, not a demonstrated mechanism.
Indoor air / Other evidence
Low-molecular-weight PAHs, dominated by naphthalene, made up roughly 87-92% of total PAH mass from scented candle combustion across all tested ventilation and humidity conditions, and were mostly in the gas phase rather than bound to particles.
Moderate evidence.
Caveats
Single scented-candle formulation, small chamber study; benzo[a]pyrene itself was non-detect in nearly every condition, so toxicity-weighted risk is driven by other high-potency congeners, not BaP.
Indoor air / Measured in people or real products
Burning four unscented paraffin candles at once drove living-room NOx and HONO to 46.86 ppb and 1.63 ppb, about ten times higher than a single lavender-scented candle (0.02 ppb NOx, 0.23 ppb HONO) burned in the same room.
Moderate evidence.
Caveats
Single home, one scented and one unscented candle product; the unscented scenario burned roughly 22x more total wax mass (4 candles vs. 1), so the room-concentration gap partly reflects source size, not just chemistry.
Source: Deng 2026, Environment International Link to this fact
Indoor air / Measured in people or real products
Once normalized per gram of wax burned, the scented candle's HONO emission rate (170.77 mg per kg of wax) was an order of magnitude higher than the unscented candles' (24.14 mg/kg) or incense sticks' (52.50 mg/kg), and higher than HONO from vehicle exhaust in a traffic tunnel (88-115 mg/kg fuel).
Moderate evidence: One scented candle product tested; emission index relies on a mass-balance model of room volume and air exchange, not direct stack sampling.
Source: Deng 2026, Environment International Link to this fact
Indoor air / Measured in people or real products
Incense smoke was far more chemically complex than candle smoke: over 100 distinct VOCs, including classic wood-smoke markers (acetonitrile, methoxyphenols) and a set of nitrogen-ring compounds (pyrroles) that were completely absent from either candle's emissions.
Moderate evidence: One incense brand vs. one scented and one unscented candle product.
Source: Deng 2026, Environment International Link to this fact
Indoor air / Measured in people or real products
Ammonia, a byproduct rarely discussed for indoor combustion, rose 3-9 ppb above background during candle and incense burning (to 30-38 ppb total) and 4-10 ppb above background during cooking (to 30-44 ppb total), in both cases a statistically significant increase over pre-combustion levels.
Moderate evidence: Single-home measurements; cooking-related ammonia rise could not be cleanly separated from concurrent human-occupancy ammonia emissions.
Source: Deng 2026, Environment International Link to this fact
Indoor air / Other evidence
Burning scented candles releases VOCs, SVOCs, and particulate matter, including polycyclic aromatic hydrocarbons (benzo[a]pyrene, naphthalene, anthracene, pyrene) classified as carcinogenic.
Moderate evidence.
Caveats
This is a secondhand restatement in a two-page opinion piece, not a new measurement; our own chamber primaries (Salthammer 2021, Andersen 2021) independently confirm PAH/VOC output but find it strongly formulation-dependent rather than universal to all scented candles.
Source: Singh 2023, Environmental Science & Technology Link to this fact
Indoor air / Other evidence
In a small case series, removing mothballs and reducing scented cleaning/personal-care products, candles, and stored chemicals cut targeted indoor-air VOC categories by roughly a quarter to over 95% (e.g., naphthalene from mothballs down 96%, p-dichlorobenzene from moth crystals down 56-80%, odorants/fragrances down 26-80%) across three homes, measured by GC/MS air sampling before and after a structured home intervention.
Moderate evidence.
Caveats
Small, uncontrolled, non-blinded case series with multiple co-occurring intervention elements per home (product substitution, education, counseling, and in one case new chemical storage); cannot isolate the VOC-reduction effect from other changes.
Source: Rincón et al. 2024, Primary Health Care Research & Development Link to this fact
Information, not medical advice. See also: myths we won’t tell you.