Field sobriety tests are less accurate than most drivers assume. NHTSA’s own validation research puts the three standardized tests at roughly 91 percent accurate when used together and administered by the book, meaning about one in ten arrest decisions based on them is wrong. Independent research suggests the real-world error rate runs higher once medical conditions, bad roadside conditions, and the subjective scoring officers do under pressure enter the picture. So how accurate are field sobriety tests in practice? Accurate enough to support probable cause for an arrest, not accurate enough to be treated as proof you were drunk.
The Accuracy Numbers by Test
The most widely cited data comes from a 1998 San Diego field study by Jack Stuster and Marcelline Burns, who rode along with officers during real stops and compared field sobriety test (FST) arrest decisions against chemical test results. For correctly classifying drivers at or above 0.08 BAC, the study found:1Office of Justice Programs. Validation of the Standardized Field Sobriety Test Battery at BACs Below 0.10 Percent
- Horizontal Gaze Nystagmus (HGN): 88 percent
- One-Leg Stand: 83 percent
- Walk-and-Turn: 79 percent
- All three combined: 91 percent
Two things about those figures. First, they were produced under study conditions, with researchers present and officers aware their work was being evaluated. Second, 91 percent accurate also means 9 percent wrong. The study authors noted the combined figure could reach 94 percent “if explanations for some of the false positives are accepted,” which is itself an admission that sober people were flagged as impaired during the research.1Office of Justice Programs. Validation of the Standardized Field Sobriety Test Battery at BACs Below 0.10 Percent
The Walk-and-Turn is the weakest of the three. At 79 percent, roughly one in five people flagged by that test alone are not actually over the legal limit. The failure threshold is also low: just two of eight possible clues is enough for a fail, and plenty of sober people display two clues on a dark shoulder at two in the morning.2National Highway Traffic Safety Administration. DWI Detection and SFST Refresher Training
What Independent Research Shows
A 1994 study published in Perceptual and Motor Skills videotaped 21 people who had been confirmed sober through breath tests performing field sobriety tasks. When certified officers later reviewed the footage, they judged 46 percent of the sober participants as too impaired to drive. Nearly half of people with a 0.00 BAC were failed by trained professionals watching the same tests NHTSA validated.
That gap, between the controlled 91 percent figure and the ugly real-world numbers, is where most of the practical story lives. It exists because the clues officers are trained to count can be produced by a long list of things that have nothing to do with alcohol.
Why Sober People Fail
Medical Conditions and Medications
HGN is particularly exposed to medical false positives. Nystagmus can be caused by inner ear disorders like benign positional vertigo, labyrinthitis, and Meniere’s disease, as well as stroke, multiple sclerosis, brain tumors, head injuries, and vitamin B12 deficiency.3MedlinePlus. Nystagmus A driver with any of these could show textbook HGN clues without having touched alcohol.
Prescription drugs add more exposure. Antiseizure medications like carbamazepine and phenytoin are among the strongest triggers for nystagmus. Antipsychotics, antidepressants including fluoxetine and sertraline, and sedatives like lorazepam can all produce eye-jerking patterns indistinguishable from those alcohol causes.4National Center for Biotechnology Information. Pharmacovigilance Study of Drug-Induced Eye Movement Disorder An officer running the test at roadside has no way to distinguish the two.
The balance-dependent tests have the same problem for people with musculoskeletal injuries, back or leg pain, inner ear dysfunction, or neurological conditions affecting coordination. NHTSA’s own training materials acknowledge these affect performance; whether an officer accounts for them is up to the officer.
Age, Weight, and Fatigue
Standing on one leg for 30 seconds is difficult for a lot of sober people. NHTSA’s validation research was done primarily on younger, relatively fit subjects, and the agency’s training materials note that people over 65 and those significantly overweight may struggle with the physical tests regardless of sobriety. Fatigue works the same way. A driver pulled over after a long shift can show diminished balance and slower reactions that look exactly like the standardized impairment clues.
The Roadside Itself
NHTSA’s manual admits the tests are designed for “ideal conditions” that “do not always exist” at roadside, and that variations like an uneven road surface “may have some effect on the evidentiary weight given to the results.”5National Highway Traffic Safety Administration. 2023 SFST Refresher Participant Manual In practice, officers administer these tests on gravel shoulders, sloped roads, in rain, under flashing emergency lights, and alongside highway traffic. Walking heel-to-toe on a highway shoulder at night is hard even when you’re stone sober.
Stress and Clothing
Being pulled over spikes adrenaline for almost anyone. Trembling hands, unsteady legs, difficulty concentrating, and shallow breathing are all common responses to the encounter itself, and they are also the symptoms officers are trained to read as impairment. Restrictive clothing and footwear add another layer: high heels, flip-flops, boots, or tight pants can physically prevent you from doing the heel-to-toe walk or one-leg stand the way the scoring demands.
What the Accuracy Figure Means for an Arrest
Field sobriety tests are not stand-alone proof of impairment. Officers combine FST observations with the reason for the stop, driving behavior, the smell of alcohol, slurred speech, bloodshot eyes, and statements the driver makes. Together, those observations build probable cause for a DUI arrest. The 91 percent figure describes the tests as a screening tool for that decision, not as evidence of a specific BAC.
The subjective scoring is the central weakness. Two officers watching the same Walk-and-Turn can disagree about whether a step missed the heel-to-toe requirement or whether the arms rose more than six inches. Each clue is a judgment call made under time pressure at the roadside.
Body camera and dash camera footage has become the main check on that subjectivity. Video captures the road surface, the lighting, how clearly the officer gave instructions, and whether the performance was really as poor as the written report describes. Reports are often written hours later; memories drift. The footage doesn’t.
In court, most jurisdictions let FST results in as circumstantial evidence of impairment, but courts universally agree the tests cannot prove a specific blood alcohol level. HGN gets extra scrutiny in some states, where it is treated as a scientific technique requiring expert testimony about the alcohol-nystagmus link before results can be admitted.
Non-Standardized Tests Have No Validation at All
If an officer asked you to recite the alphabet, count backward, touch your finger to your nose with your eyes closed, or stand with feet together and eyes closed while estimating 30 seconds, those are not NHTSA-validated tests. The accuracy numbers above do not apply to them. NHTSA does not recognize any of these as reliable indicators of impairment, and they have never been through the validation process that produced the figures for HGN, Walk-and-Turn, and One-Leg Stand. Their evidentiary weight is weaker and they are easier to challenge.
If You’re Pulled Over
Field sobriety tests are voluntary in most jurisdictions. You can politely decline them without a legal penalty for the refusal itself. Many drivers don’t know that, and assume they have to comply with whatever the officer asks.
Declining doesn’t end the encounter. If the officer has other reasons to suspect impairment, those alone can supply probable cause for an arrest, after which you may be asked to take a chemical test. Chemical tests are a different legal animal: every state has an implied consent law, and refusing a chemical test after a lawful arrest triggers automatic administrative penalties, typically a license suspension of six months to a year, and the refusal can be used against you at trial. The U.S. Supreme Court held in Birchfield v. North Dakota that states can require a warrantless breath test incident to a lawful DUI arrest but generally need a warrant for a blood draw, and cannot impose criminal penalties for refusing the blood test.6Justia. Birchfield v North Dakota
Because FSTs rely so heavily on officer judgment and controlled administration, they create multiple openings for a defense. Improper administration is the most common: skipped instructions, an HGN stimulus moved too fast, no inquiry about medical conditions. Environmental and physical factors are the second: sloped surfaces, rain, distracting lights, unsuitable footwear, a medical condition or prescription that produces the exact clues being counted. The third is the stop itself: if the officer lacked reasonable suspicion to pull you over or probable cause to arrest, a motion to suppress can exclude the FST results along with everything else that flowed from the stop.
The practical bottom line on accuracy: a 91 percent screening tool is good enough to justify a closer look, not good enough to settle the question. Treat an FST result as the beginning of the evidence against you, not the end of it, and if you’ve been arrested based on one, the specifics of how the test was given, where, and in what condition you were in when you took it are worth going through carefully with a defense attorney.