The claim that runs through a new national investigation is blunt: the roadside sobriety tests police use to build drunk-driving cases "catch everybody," sober and impaired alike. That phrase anchors a July 9, 2026 WSMV4 report out of Nashville, the culmination of a multi-year series that examined cases across 22 states. The reporting draws a distinction that defense lawyers have argued for years and that courts are now taking seriously: the three standardized field sobriety tests may be effective at generating probable cause to arrest, yet they are not accurate enough to stand alone as courtroom proof that a driver was impaired.
That gap between probable cause and proof beyond a reasonable doubt is the whole story. It is also the difference between an arrest that is lawful and a conviction that is sound. If you are facing a DUI or DWI charge that rests on how you performed at the side of the road, understanding that distinction is where a defense begins.
What the three tests actually measure
The National Highway Traffic Safety Administration (NHTSA) standardized three tests, collectively the Standardized Field Sobriety Tests (SFSTs), from research the agency commissioned at the Southern California Research Institute beginning in the mid-1970s, with foundational studies in 1977 and 1981. The three tests are:
-
Horizontal gaze nystagmus (HGN). The officer moves a stimulus, often a pen or fingertip, across your field of vision and watches your eyes. Nystagmus is an involuntary jerking of the eye. The officer counts "clues" such as the eye jerking before it reaches a certain angle.
-
Walk-and-turn. A "divided attention" test. You take nine heel-to-toe steps along a line, turn in a prescribed way, and walk back while following instructions. Stepping off the line, starting too soon, or using your arms for balance each count as clues.
-
One-leg stand. A balance test. You stand on one foot, the other raised roughly six inches, and count aloud. Swaying, hopping, or putting your foot down are scored as clues.
On each test, a "clue" is a specific scored behavior, and hitting a set number of clues is what the officer treats as a "fail." The scoring is meant to be objective. In practice, the number of clues an officer records can decide whether you are arrested.
Where the reliability numbers come from
NHTSA markets the full battery as roughly 91% reliable at detecting a blood alcohol concentration (BAC) at or above 0.08%, with the walk-and-turn cited at about 79% and the one-leg stand at about 83% individually when a driver shows two or more clues. Those figures come largely from a 1998 field validation study of 297 drivers in San Diego. When an arresting officer testifies about how accurate the tests are, this is often the study behind the number.
The problem, critics say, is what those headline percentages leave out. A detailed 2021 critique in the Oklahoma Bar Journal re-examined the same 1998 San Diego data and found substantial false-positive exposure among drivers who were below the legal limit. Among sub-0.08% drivers, roughly 37% still showed four or more HGN cues, about 53% showed two or more walk-and-turn clues, and about 41% showed two or more one-leg-stand clues. The critique also notes that the study sample skewed heavily male (about 12% female) with a high mean BAC of about 0.122%, which limits how well its results describe borderline or sober drivers, the very people most likely to contest an arrest.
The false-positive problem, in a controlled trial
The strongest peer-reviewed support for the investigation's central claim is a 2023 randomized clinical trial published in JAMA Psychiatry. Researchers at the University of California San Diego Center for Medicinal Cannabis Research (Marcotte and colleagues) ran a double-blind, placebo-controlled study of 184 cannabis users. About 70 minutes after dosing, trained officers evaluated participants using the standardized tests.
Officers classified 81.0% of the THC group as impaired. They also classified 49.2% of the sober placebo group as impaired. In other words, nearly half of people who had received no active drug were rated as impaired on the roadside tests. The authors concluded that the SFSTs are not accurate enough on their own to establish cannabis impairment.
That result did not appear out of nowhere. As early as 1994, research by Spurgeon Cole found that examiners rated a substantial share of subjects with a 0.00% BAC as "too impaired to drive," an early demonstration of the same false-positive pattern.
Why do sober people fail? The reasons are ordinary and well documented in defense practice. Some people have naturally occurring nystagmus unrelated to alcohol. Neurological and inner-ear conditions, certain medications, age, recent injuries, and fatigue can all affect balance and eye movement. So can the setting: the slope or "grade" of the road, gravel or uneven pavement, poor lighting, cold or wind, the shoes you happen to be wearing, and plain nervousness at being stopped by police at night. None of those has anything to do with intoxication, yet each can produce the clues an officer is trained to count.
The courtroom split, part one: HGN as expert testimony
Courts are divided on how much weight this evidence can carry, and the fault line often runs through the HGN test, because reading involuntary eye movement is closer to a medical or scientific opinion than to a lay observation.
A recent example is State v. Alcenat, decided by the Connecticut Appellate Court on July 14, 2026. The defendant, Charles Alcenat, appealed an operating-under-the-influence conviction under Connecticut General Statutes § 14-227a. He argued that the trial court abused its discretion by allowing the arresting officer, identified as Murphy, to testify as an "expert" on HGN, testimony the defense said made up "almost the entirety of evidence of alcohol consumption." The appellate court reviewed the admission of that HGN-expert testimony under Connecticut's abuse-of-discretion and harmless-error framework, a posture in which a non-constitutional evidentiary error places the burden on the defendant to show that the error caused harm.
Alcenat illustrates a national reality rather than a national rule. Some courts admit HGN evidence only for the fact of impairment and not as a correlation to a specific BAC number. Others exclude officer HGN "expert" testimony unless the state lays a scientific predicate first. The trend is more litigation over admissibility, not a single settled standard.
The courtroom split, part two: drug recognition experts
The same scrutiny has reached Drug Recognition Expert (DRE) evaluations, the multi-step protocol officers use to identify drug impairment when alcohol is not the issue. The leading example of a state high court examining that science is State v. Olenowski, decided by the New Jersey Supreme Court on November 15, 2023, by a 5-2 vote.
After evaluating the DRE methodology under a Daubert-style reliability standard, the court held that DRE testimony is admissible, but it cabined that testimony in two important ways. A DRE may testify only that findings are "consistent with" the ingestion of a drug, not that drugs caused impairment. And DRE testimony must be excluded where the state fails, without persuasive justification, to reasonably attempt to obtain a toxicology report. The New Jersey Monitor summarized the ruling as deeming drug recognition experts "reliable enough," a phrase that captures both halves of the holding: reliable enough to be heard, but not reliable enough to go unchecked.
How the defense fights back
The reporting and the case law point to the same set of defense moves, all built on separating probable cause from proof.
-
Pretrial motions to exclude or limit the testimony. Defense counsel challenge the foundation for the evidence, the officer's qualifications to give it, and any deviation from NHTSA's own administration protocol. Tests given differently than the standardized instructions are, by NHTSA's own logic, less reliable.
-
Reliability challenges. Depending on the jurisdiction, lawyers raise Daubert, Frye, or a state-equivalent standard to question whether officer testimony about HGN or DRE findings rests on accepted science.
-
Cross-examination on non-alcohol causes. This is where the false-positive research does its work. Counsel walk through the medical, physical, and environmental explanations for each clue: natural nystagmus, neurological or inner-ear conditions, age, injury, footwear, road grade, weather, and anxiety.
The unifying theme is the standard of proof. An officer can reasonably believe you are impaired, and be wrong, and still have had probable cause to arrest. Conviction requires more. When the roadside tests are the bulk of the state's case, the defense argues that "good enough to arrest" is not "proof beyond a reasonable doubt."
The trend, and what is not yet settled
The direction of travel is toward more admissibility challenges, more expert testimony on both sides, and louder calls for disclosure of false-positive rates and for better officer training. The WSMV4 experts, including forensic toxicologist Dr. Stefan Rose, DUI defense attorney Carlos Canet, and former officer and DRE specialist Joshua Ott, frame the tests the same way: effective at generating probable cause, but not accurate as proof of impairment.
A caution is in order. Field sobriety tests remain widely admitted in courts across the country. The 49.2% figure comes from a cannabis study, and the reliability debate is contested, not resolved. What has changed is that judges are increasingly willing to ask how the tests were administered, what the science actually shows, and whether an officer should be allowed to present roadside observations as expert conclusions. For a driver whose case turns on a few clues counted at the roadside, that shift matters.
Related reading
-
[Half of Sober Drivers Fail the Roadside Marijuana Test: Why Courts Are Starting to Throw Out Cannabis DUI Field Evidence](/article/sober-drivers-fail-roadside-marijuana-test-cannabis-dui-field-evidence)
-
In Mississippi, a Gun Doesn't Have to Be Fired, or Even Used, to Send a 13-Year-Old to Adult Court
Sources
-
Sober drivers arrested for DUI in 22 states, WSMV4 investigation finds
-
State v. Alcenat (Connecticut Appellate Court, July 14, 2026, slip opinion)
-
New Jersey's top court deems drug recognition experts 'reliable enough' (New Jersey Monitor)
-
N.J. high court weighs science behind drug recognition experts (WHYY)
Note: This article contains AI-assisted content and has been reviewed by our editorial team.
IMPORTANT NOTICE: GavelDaily.com is an advertising and information service, not a law firm. Submitting this form does not create an attorney-client relationship. The information you provide will be shared with licensed criminal defense attorneys in your area who may contact you about your inquiry. All submissions are treated as confidential by our service, but attorney-client privilege does not attach until you formally retain an attorney. Any attorney you connect with through this service is independently responsible for their legal advice and representation. Free consultations are subject to individual attorney availability. If you are facing an immediate legal emergency, contact your local public defender's office or call 911.