Does High-Sulphite Wine Change Driving at the Same Alcohol Dose? A 2026 Simulator Study

Does wine with more sulphites make someone a more dangerous driver at the same alcohol dose? One controlled driving-simulator experiment found some differences in simulated urban driving after participants drank higher-sulphite wine, but the answer is not established for an individual driver. The experiment was small, outcomes varied by road task, and several reported statistical comparisons were against a sober baseline rather than a direct high-versus-low sulphite contrast. Investigators measured wine composition, obtained breathalyser-based alcohol readings, and recorded simulator behavior. They did not directly assay venous whole-blood ethanol, observe real crashes, validate a sulphite-specific impairment test, or determine the cause of any particular collision.

The original study is Bassani and colleagues, “Understanding driver behaviour after consuming conventional wines with different sulphite contents: Insights from a driving simulator study”, PLOS One (2026). Its narrow findings can inform questions about experimental driving behavior, but they must not be converted into a universal conclusion about wine additives, BAC, or legal impairment.

What exactly was tested?

The researchers used a triple-blind, randomized, within-participant design. Thirty-two licensed adults (16 women and 16 men) each drove a fixed-base simulator in three sessions on different days: sober, after a lower-sulphite red wine, and after a higher-sulphite red wine. The order was randomized, and sessions were separated by at least a week. Participants were selected from a restricted population; the study excluded abstainers and people drinking more than four alcohol units weekly, among other criteria. Those selection rules limit how directly the results apply to the broader population of drivers.

Both experimental wines were produced from Barbera grapes and had an ethanol content of 15.2% by volume. Measured total sulphur dioxide/sulphite content was 86 mg/L in the lower-sulphite wine and 126 mg/L in the higher-sulphite wine. On the drinking visits, volunteers received 190 mL of wine, which the authors describe as approximately 1.5 alcohol units. The amount was selected to target an alcohol concentration around 0.5 per mille, the general Italian driving limit discussed in the paper; it was not a guarantee that each participant had exactly that concentration.

The fixed-base simulator presented rural roadway curves, a rural-to-urban speed transition, sudden pedestrian appearances at crossings, and sustained following of a slower vehicle. Each simulated driving session lasted about ten minutes. These are specific laboratory tasks, not a sample of normal travel over many hours and not an observed crash investigation.

What was measured directly, and what was inferred?

The study reports “BAC” measurements before alcohol, approximately 20 minutes after wine, and at the end of driving. The Methods expressly describe a breathalyser; they do not describe collection and gas-chromatographic quantification of venous or arterial blood. The resulting “BAC” notation should therefore be treated as a breath-instrument-derived blood-alcohol-equivalent estimate unless original analytical records establish otherwise. A numerical label on an instrument or in a data table does not transform an exhaled-breath measurement into a direct blood-specimen measurement.

Likewise, sulphite concentration was characterized in the administered wine, not established as an individually quantified sulphite concentration in participants’ blood. Steering, lane placement, speed, headway, and surrogate conflict-time variables came from the simulator. Linear mixed-effects models supplied adjusted associations. Those model coefficients are not independent biochemical assays, personal impairment scores, or proof of biological mechanism.

For background on the analytical distinction, see this site’s guides to breath alcohol measurement and pharmacokinetics in toxicology.

What happened during rural driving?

In the relatively uncomplicated rural simulation, the investigators found no significant effect attributed to the difference in wine sulphite content on the analyzed lane-control and speed measures. One key association concerned the end-of-session alcohol reading and lane-position variability during curves: higher reported alcohol values were associated with more lateral variation (p = .004). The authors note that alcohol accounted for only a modest portion of variance in that model; the modeled driver-specific component remained important.

That result concerns alcohol-related lane-position variability in a simulator, not a demonstrated sulphite effect. The study also did not record participants leaving the lane or losing vehicle control in that rural section. Absence of such events in this short and low-demand experiment is not a validated finding that drinking near a particular legal threshold is safe.

What did the simulated pedestrian event show?

In the urban crossing task, the primary measure was minimum time-to-collision (MTTC): a modeled minimum time margin between road users if their speed and direction were unchanged. Against the sober reference condition, the higher-sulphite treatment was associated with an approximately 2.5-second increase in MTTC (p = .015). The lower-sulphite estimate was around two seconds, with p = .051, slightly above the conventional .05 cutoff. The model also associated higher end-of-session alcohol estimates with lower MTTC. A second conflict measure, post-encroachment time (PET), was associated with the alcohol variable (p = .034) but not significantly with sulphite treatment.

A longer modeled margin at a particular crossing is one potentially more cautious observed response. It does not establish that higher sulphites improve driving safety. The investigators did not directly measure perceived risk, workload, or the proposed compensatory behavioral mechanism. Nor does a statistically significant high-sulphite-versus-sober coefficient, by itself, prove a statistically significant high-sulphite-versus-low-sulphite difference.

What happened when drivers followed another vehicle?

In the continuous urban car-following task, the higher-sulphite condition was associated with higher maximum speed (p = .010), greater speed variability (p = .013), and a shorter minimum time headway (p = .042) compared with the sober reference in the relevant models. A shorter headway reduces the modeled temporal spacing behind the lead vehicle; this is a task-specific surrogate for following distance or safety margin, not an observed rear-end crash.

Separately, the end-of-session alcohol estimate was associated with increased lateral-position variability in urban driving (p = .006). This is an alcohol-associated lane-control measure, not a direct sulphite assay. Other factors, including participant-level differences, accounted for substantial portions of the observed variability.

The apparent contrast between a larger margin during one pedestrian encounter and less favorable car-following measurements is not necessarily contradictory: a brief unexpected conflict and sustained regulation behind a lead vehicle involve different tasks. Nevertheless, the results do not validate a mechanistic explanation, do not provide a diagnostic cutoff, and should not be generalized beyond the experimental conditions.

Why the comparator matters more than a dramatic headline

A reader may interpret “the high-sulphite wine produced a significant effect” to mean researchers conclusively showed the high-sulphite wine differed from the low-sulphite wine. But the study’s models used the sober condition as a reference for several reported treatment coefficients. A statistically significant coefficient for the high-sulphite condition and a nonsignificant one for the low-sulphite condition are not, by themselves, a significance test of the difference between those two coefficients.

Because both wine conditions involved ethanol and differed from sober baseline, a clean sulphite-specific claim requires the relevant direct between-wine contrast, with its estimate and uncertainty, not an inference assembled from separate baseline comparisons. The paper supports a cautious description of condition- and context-dependent simulator associations. It should not be cited as proof that a particular sulphite concentration independently caused a particular driver’s bad behavior.

Key findings and limits at a glance

Can the results decide whether an individual driver was impaired?

No. Population-level task measurements and adjusted coefficients do not establish that a named driver violated a statutory standard, was unable to drive safely, or caused a collision. Actual case interpretation could require a properly identified and time-stamped biological specimen, validated analytical method, relevant legal jurisdiction, observations of driving and behavior, road conditions, alternative explanations, and temporal linkage between events and any alcohol measurement.

The experiment compared a restricted group of 32 volunteers, not a validation population for impairment classification. It did not establish sensitivity, specificity, individual prediction limits, an impairment threshold for wine sulphites, or a method for converting a breath instrument result into a measured blood concentration. It examined simulated driving over a short period, with only selected adverse and protective indicators showing significance. The authors acknowledge possible simulation-related discomfort and the absence of systematic symptom assessment.

Moreover, a legal limit is a regulatory boundary—not an experimentally validated guarantee of no driving-performance effects below that boundary. Italian regulatory context cannot simply be transferred to another jurisdiction. For a general scientific framework on what toxicological results can and cannot establish, consult the site’s forensic toxicology evidence library.

Frequently asked forensic questions

Did researchers directly measure blood alcohol concentration?

The Methods describe breathalyser testing at defined times, even though the article labels reported readings “BAC.” No paired quantitative venous whole-blood alcohol assay is described. Distinguish the actual breath method from its blood-equivalent reporting convention.

Were the two wines identical except for sulphites?

The researchers deliberately prepared two conventional Barbera wines with different total sulphur dioxide content and reported the same nominal 15.2% alcohol strength. That experimental control improves comparison, but does not prove that every potentially relevant chemical or perceptual feature was identical or that the biological mechanism of any driving difference was sulphite-specific.

Does a significant result for high sulphites mean high versus low wines differed significantly?

Not automatically. Separate high-versus-sober and low-versus-sober tests cannot substitute for a direct high-versus-low statistical contrast. Inspect the exact coefficient, reference group, adjusted comparison and uncertainty before making that claim.

Did sulphites improve pedestrian safety?

In one simulated crossing task, the high-sulphite treatment corresponded to a larger modeled minimum time-to-collision compared with sober baseline. This does not demonstrate improved overall safety, a lower real crash rate, or a direct benefit over the low-sulphite wine.

Did higher sulphites worsen all driving outcomes?

No. Some rural and transition outcomes did not differ by sulphite treatment. Adverse associations appeared in selected urban car-following measures, while one pedestrian margin moved in the opposite direction. The effect pattern depended on the task and comparison.

Can this study establish a sulphite intoxication cutoff?

No. The 86 and 126 mg/L values describe the administered beverages. The researchers did not measure circulating sulphite levels, define a personal dose-response threshold, or validate an intoxication classifier.

Does driving below a country’s legal BAC limit establish absence of impairment?

No. Statutory limits are legal rules, not categorical clinical guarantees. The study’s own models associated the breath-derived alcohol variable with selected simulator outcomes. An individual impairment opinion requires evidence beyond whether an estimated concentration was below one jurisdiction’s limit.

Can the experiment determine that wine additives caused a real accident?

No. There were no real traffic crashes in the experiment. Model coefficients from simulator tasks cannot establish the biological and legal causation of a particular collision.

Primary scientific source

Bassani M, De Paolis C, Gerbi V, Giacosa S, Hazoor A, Lioi A, et al. (2026). Understanding driver behaviour after consuming conventional wines with different sulphite contents: Insights from a driving simulator study. PLOS One 21(7):e0345134. doi:10.1371/journal.pone.0345134. The interpretation here is confined to the study’s measured beverage composition, breath-based alcohol measurements, simulated tasks, model associations, and stated limitations. It does not constitute an opinion in any individual case.