Every standardized field sobriety exercise rests on a small set of studies from the National Highway Traffic Safety Administration, and the single rule those studies produced is the one that matters most in your case. NHTSA states that the exercises are only valid when they are administered in the prescribed, standardized manner, and that if any one of the standardized elements is changed, the validity is compromised. That sentence is the cardinal rule, and it is the foundation of the defense.
The research itself came out of the Southern California Research Institute, beginning with Burns and Moskowitz in 1977, followed by Tharp, Burns, and Moskowitz in 1981, and Anderson, Schweitz, and Snyder in 1983. Dr. Marcelline Burns later supervised a Florida validation study, working with the Pinellas County Sheriff’s Office. A 1998 San Diego study reported higher accuracy figures, and the State leans on it, but every one of these studies measured performance under controlled conditions by trained administrators following the procedure exactly.
The original NHTSA validation accuracy, even when the exercises were administered correctly. Each one leaves a sizable share of people misclassified, which is the opposite of certainty.
The numbers were never proof beyond a reasonable doubt
Even taken at face value, the original validation classified horizontal gaze nystagmus correctly only about 77 percent of the time, the walk and turn about 68 percent, and the one leg stand about 65 percent. Read those numbers the other way and the walk and turn misclassified roughly a third of people, and the one leg stand more than a third. A later review by McKnight and colleagues in 2002 found that at low blood alcohol levels the only index that held up at all was horizontal gaze nystagmus, and even it produced a false positive in roughly 38 percent of cases at the standard cutoff, while the balance exercises did not reliably separate people at all.
What the studies measured, and what they did not
There is a gap in these studies that the accuracy figures hide. The validation research asked one narrow question, which is whether the exercises predict a blood alcohol number at or above a legal threshold, tested on volunteers under controlled conditions by administrators who followed the procedure exactly. The studies did not measure whether a person was driving unsafely, and they were never built to. A DUI charge under section 316.193, Florida Statutes, turns on impairment of normal faculties or an unlawful blood or breath level, so a roadside exercise that lines up only loosely with a laboratory reading is answering a different question than the one the jury has to decide. That distinction is easy to lose in a report full of clues and hard to ignore once someone who teaches the battery lays it out.
| Study | What it measured | Headline figure | The catch |
|---|---|---|---|
| Burns and Moskowitz, 1977 | The lab basis for the battery | More than 83 percent classified by BAC | A controlled lab, trained raters, and a BAC threshold rather than driving |
| Tharp, Burns, and Moskowitz, 1981 | Accuracy of each exercise | HGN 77, walk and turn 68, one leg stand 65 percent; the two together 92 percent | A third or more of people misclassified on the balance exercises |
| Colorado, 1995 | Officer arrest and release decisions in the field | 86 percent of decisions correct, 93 percent of arrests correct | Scores the officer’s whole decision, not the exercises alone |
| Florida, 1997 | Officer arrest decisions, Burns with the Pinellas County Sheriff’s Office | 95 percent of arrest decisions correct, 82 percent of releases correct | The same field-decision caveat, under ideal study conditions |
| San Diego, 1998 | Officer decisions at BACs below 0.10 | 91 percent of combined decisions correct at 0.08, up to 94 with explanations | Used modified scoring and seven hand-picked officers, so not the standard battery |
| McKnight and colleagues, 2002 | Performance at low BAC | Only HGN held up, with about 38 percent false positives at the cutoff | The balance exercises did not reliably separate people |
Every row here was produced under controlled conditions by administrators following the procedure exactly. The roadside is not that setting, which is where the defense begins.
The cardinal rule, and the warning NHTSA stopped printing
For years the NHTSA manual printed a bold warning that any deviation from the standardized procedures could compromise the results. Starting with the 2004 edition, that bold warning was removed from the manual, even though the underlying science did not change. The exercises are still only valid when administered as trained. As an NHTSA-qualified instructor, I can put the older edition and the current one side by side and show that the requirement of strict, standardized administration is exactly what the training has always said.
The training standard behind the badge
The same body that wrote the procedure also set the training standard, and it asks for more than many arrests reflect. The IACP and NHTSA model calls for a 24-hour practitioner course, a written exam passed at 80 percent or better, and a refresher recommended at least every three years. When an officer’s certification has lapsed, or the file does not show current training, that is a foundation problem before the first clue is ever discussed. Because I hold the field sobriety instructor credential and have taught this course, I can read a training record the way the standard intends and show a jury where the officer’s own qualifications fall short of what the battery requires.
How Florida treats a deviation
Florida is not a strict-compliance state. Unlike Ohio, where State v. Homan, 732 N.E.2d 952 (Ohio 2000), once required strict compliance for admissibility, Florida generally lets the exercises in even when the officer departed from the procedure, and treats the deviation as a question of weight for the jury rather than a reason to keep the evidence out. That does not weaken the defense, it locates it. The fight happens at cross-examination and in argument, where a deviation from an NHTSA standard, explained by someone who teaches that standard, becomes reasonable doubt the jury can hold onto.
You are not a number on a printout to me, and you will not be one in that courtroom either. When I take one of these cases, I put the government’s own studies and its own manuals in front of the jury and show them, in plain terms, every place the roadside fell short of what the science requires. I teach this battery, so I can do that from the standard itself, calmly and without theatrics, until the people in that box understand that a set of numbers built in a lab has almost nothing to do with what happened to you on the side of the road.
I started out as an Assistant Public Defender in Florida’s Thirteenth Judicial Circuit, in Tampa, and today I am one of six ACS-CHAL Forensic Lawyer-Scientists in Florida and an NHTSA-qualified field sobriety instructor. I was trained to administer this battery and to teach the officers who give it, so I can stand in front of a jury and show the exact point where the roadside exercises in your case left the standard. Learn more about my background.
Where the Standardization Numbers Come From
The battery was developed in laboratory research in the late 1970s and then field tested in a series of state studies. In the Colorado study, officers using the battery made correct decisions about 86 percent of the time overall at the state’s 0.05 threshold, but only 64 percent of their release decisions were correct. The 1998 San Diego study published 91 percent overall accuracy at 0.08, and it is the study officers are taught to cite.
Here is what the citation leaves out. The study used a modified scoring procedure and seven officers from a single alcohol enforcement unit, and the drivers tested averaged 0.122 while those released averaged under 0.05, which means it graded the battery mostly on decisions that were already obvious. Among the drivers who were under 0.08, roughly 29 percent were still judged over, and reading the same data to the end makes the picture the State prefers harder to hold. Four or more HGN clues showed up in about 37 percent of those sober drivers, two or more walk and turn clues in about 52 percent, and two or more one leg stand clues in about 41 percent. On the walk and turn, a sober person was about as likely to be marked as failing as an officer would be flipping a coin. Sixty-seven percent of the people wrongly arrested under the limit had all six HGN clues, so a perfect eye score did not keep an innocent person out of handcuffs. None of that comes from the defense. It is the government’s own study, read all the way through.
Questions About the NHTSA Standard
What is the NHTSA standard for field sobriety tests?
It is the fixed procedure for giving and scoring each standardized exercise. NHTSA states the exercises are only valid when administered in the prescribed, standardized way, and that changing any element compromises the validity. That rule is the heart of the defense.
How accurate are the standardized tests?
The original validation found horizontal gaze nystagmus correct about 77 percent of the time, the walk and turn about 68 percent, and the one leg stand about 65 percent, and only under controlled conditions. That leaves a large share of people misclassified, which is far from certainty.
What did the McKnight study find?
A 2002 review found that at low blood alcohol levels only horizontal gaze nystagmus held up at all, and even it gave a false positive in roughly 38 percent of cases at the standard cutoff, while the balance exercises did not reliably separate people.
Did NHTSA remove a warning from its manual?
Yes. Earlier editions printed a bold warning that any deviation could compromise the results. Starting with the 2004 edition that bold warning was dropped, even though the science did not change and the exercises are still only valid when administered as trained.
Does a deviation get the test thrown out in Florida?
Usually not on its own. Florida is not a strict-compliance state like Ohio under State v. Homan, so the exercises generally come in and the deviation goes to weight. The challenge is made at cross-examination and in argument rather than only in a pretrial motion.
Why does your training matter for this?
Because I teach this battery. As an NHTSA-qualified instructor I can explain to a jury exactly what the standard requires, show the edition history, and connect each deviation in your case to the validation conditions it broke.
Related: the three standardized tests, administration errors, HGN as scientific evidence, how clues are scored, and the field sobriety overview.
Do the field studies prove the tests work?
Not for your stop. The Colorado and Florida field studies scored the officer’s whole decision, which folds in the driving, the odor, and the statements, and they used trained administrators under study conditions. That is not the roadside, so a high field number is not proof the exercises were reliable in your case.
This page is general information, not legal advice, and it does not create an attorney-client relationship. Field sobriety exercises in Florida are governed by case law and by section 316.193, Florida Statutes. Procedures and rules change, and every case turns on its own facts. Past results do not guarantee a similar outcome.
