Were Your Field Sobriety Tests Correct? The NHTSA Numbers Nobody Shows You

It's 2 AM. You're standing on the gravel shoulder of a road you've driven a hundred times. There's a flashlight in your face. The pavement behind you is tilted a few degrees toward the ditch because that's how shoulders are built, to drain water. You're wearing the shoes you put on that morning for work, not the shoes you'd have picked if you knew you'd be asked to walk heel-to-toe on a slope in the dark.

The officer tells you to take nine steps.

You ask about the surface. He writes down "uncooperative."

Here's what nobody tells you: the test you just took has a documented accuracy rate. NHTSA, the same federal agency that trains officers to give these tests, published it. And the number is not what you think it is.

The Numbers NHTSA Published About Its Own Tests

You're in the courtroom and the officer is testifying that you "failed" the field sobriety tests. He says it with certainty. The jury hears "failed" and pictures a pass/fail exam with clear right answers. Nobody mentions that NHTSA's own research says the test is wrong a third of the time.

The three standardized field sobriety tests did not come from nowhere. In the 1970s and 1980s, NHTSA commissioned research to figure out which roadside behaviors actually correlated with blood alcohol content above legal limits. They validated three tests and threw out everything else.

Here's what the validation actually showed.

Horizontal Gaze Nystagmus (HGN). The officer moves a pen or finger across your field of view and watches your eyes. In NHTSA's validation studies, HGN was approximately 77% accurate at detecting a BAC of 0.10 or higher (NHTSA HS 178 R2/06). That means roughly 23 out of every 100 sober or borderline people would fail it under study conditions.

Walk-and-Turn. The nine-step heel-to-toe test. Approximately 68% accurate in NHTSA's validation (NHTSA HS 178 R2/06). Roughly 32 out of 100 people fail it even when they shouldn't.

One-Leg Stand. Raise one foot six inches off the ground and count. Approximately 65% accurate (NHTSA HS 178 R2/06). Roughly 35 out of 100 false results (NHTSA).

Combined battery. When officers administer all three and look at the results together, NHTSA reports around 91% accuracy (NHTSA), but only when every single protocol is followed exactly, and only under the conditions NHTSA validated.

The One-Leg Stand is wrong 35 times out of 100 under perfect conditions, and your stop was not perfect conditions.

But those conditions were not a gravel shoulder in the dark. The 91% combined number is a ceiling, not a floor. Every deviation from protocol, every environmental factor outside spec, pushes that number down, and nobody publishes how far.

Those conditions were not a gravel shoulder in the dark.

Your 5-minute action: write down which tests you were given. For each one, write down the surface you were standing on, the weather, the lighting, and what shoes you were wearing. That is the beginning of comparing your stop to the conditions NHTSA actually validated.

What NHTSA's Own Manual Says About When the Tests Don't Work

You're reading the NHTSA training manual, the same manual the officer studied. Page after page of instructions. And then you hit the section that lists all the conditions under which the tests are no longer valid. It's right there in the training materials. In plain language.

Here's the dirty truth about field sobriety tests: the NHTSA training manual, the same manual officers are required to study, lists the conditions under which the tests are no longer valid. Not as a defense trick. As part of the training.

Read the manual and you find, in plain language, that the tests were validated under specific conditions. Walk away from those conditions and the accuracy numbers do not carry over.

The manual flags:

• Uneven, sloped, or slippery surfaces. The Walk-and-Turn and One-Leg Stand were validated on flat, dry, hard, non-slippery surfaces. Gravel, grass, slopes, wet pavement, all outside the validation envelope.
• Inadequate lighting. Officers are supposed to have enough light for you to see the line. "Enough" is subjective, but headlights pointed at you from behind are not what NHTSA had in mind.
• Wind, rain, and cold. Weather affects balance. Cold affects muscle control. NHTSA did not validate these tests in a storm.
• Age 65 and older. The manual specifically notes that subjects over 65 may have difficulty with the tests for reasons unrelated to alcohol.
• 50+ pounds overweight. The One-Leg Stand in particular has known issues for subjects carrying significant extra weight.
• Back, leg, or inner ear problems. Any physical condition affecting balance compromises the Walk-and-Turn and One-Leg Stand.
• Footwear. Heels taller than two inches should be removed. The manual is explicit about this.

Every condition on this list is from NHTSA's own training manual, the document the officer was required to study.

But here's the part that matters: any single one of these puts a test outside the conditions NHTSA validated. On a 2 AM roadside stop, you might hit four or five at once. The accuracy numbers NHTSA published don't apply. And nobody tells you that.

Any one of these puts a test outside the conditions NHTSA validated. On a 2 AM roadside stop, you might hit four or five at once.

Your 5-minute action: go through the list above and check every condition that applied to your stop. Write them down. For each one, note whether the officer acknowledged it, adjusted for it, or ignored it. That checklist is what gets compared against the NHTSA manual.

The Protocol Side Nobody Talks About

The officer tells you to walk heel-to-toe. He demonstrates, sort of. He gives instructions, sort of. You start walking before he says "begin" because you thought he was done talking. He writes down "started too early" as a clue.

Environmental conditions are half the story. The other half is whether the officer administered the test the way NHTSA trained them to.

The Walk-and-Turn has specific instructions. The officer is supposed to demonstrate the steps. They are supposed to give specific verbal instructions in a specific order. They are supposed to have you stand in a specific starting position before they tell you to begin. The test has two phases, the instruction phase and the walking phase, and NHTSA validated the results based on observing both.

Miss a step in the demonstration, skip a line of instructions, let the subject start before saying "begin," and you have deviated from the protocol NHTSA validated.

The HGN has even more moving parts. There's a specific speed the stimulus is supposed to move. A specific distance from the face. A specific number of passes. A specific order of observations, checking for lack of smooth pursuit, then distinct nystagmus at maximum deviation, then onset of nystagmus before 45 degrees. Skip any of those and you're running a different test than the one that was validated at 77%.

An officer who skips a step in the NHTSA protocol is not running the test that was validated, they are running a different test with unknown accuracy.

But protocol deviations are invisible unless someone compares the dashcam to the manual, step by step. The officer's report will say "administered per training." The video may show otherwise.

Here's what nobody tells you: NHTSA studied officers under ideal conditions, with ideal training, following every step. Real traffic stops are not ideal. Dashcam reviews routinely show protocol deviations that would have invalidated the results in the original studies.

Your 5-minute action: write down everything you remember about how the officer gave you the tests. Did they demonstrate? Did they give full instructions? Did they tell you when to start? Did they hold the pen at the right distance from your face? Your memory plus the dashcam footage is the comparison set.

What This Looks Like in Practice

You're looking at the arrest report. It says you failed the Walk-and-Turn. It lists four clues. It does not mention the gravel, the slope, the wind, the cold, or your reconstructed knee.

Say you were stopped at 2 AM on a rural road. The officer had you perform the Walk-and-Turn on the shoulder. The shoulder was gravel with a slight slope toward the ditch. Your headlights were on behind you. It was 40 degrees and breezy. You were wearing work boots. You're 52, and your right knee has been reconstructed from a work injury three years ago.

Run that against the NHTSA manual.

Surface: non-compliant (gravel, sloped). Lighting: questionable (headlights from behind are not ideal illumination of the walking line). Weather: cold and windy, outside validation. Footwear: work boots are not athletic shoes and may affect gait. Physical condition: reconstructed knee affects balance for reasons unrelated to alcohol. Age: under 65, so that factor doesn't apply.

That's four environmental or physical factors outside NHTSA's validated conditions. The 68% accuracy number NHTSA published for the Walk-and-Turn does not cover this test. What the actual accuracy was for your specific stop, on your specific surface, in your specific condition, nobody knows, because NHTSA never validated it there.

Four factors outside NHTSA's validation envelope means the published accuracy numbers do not apply to your test, period.

That's not a defense argument. That's just what the NHTSA manual says.

Now change one variable. Say the stop was in a well-lit parking lot, flat asphalt, 70 degrees, no wind, and you were wearing sneakers. The only factor out of spec is your knee. That's a narrower challenge, focused on the physical condition, but still a documented deviation from the conditions under which NHTSA validated the test.

Same test. Same charge. Completely different set of questions to ask.

Your 5-minute action: take the conditions you wrote down earlier and count how many fall outside the NHTSA manual's list. Write that number at the top of the page and circle it. That is the number of reasons the published accuracy rate does not apply to your stop.

The Questions That Actually Matter

Once you know what the NHTSA manual says, these are the questions that separate a defendable FST result from one that meets the validation standard:

1. What was the surface? Flat, hard, dry, non-slippery, or something else?
2. What was the lighting? Sufficient for the subject to see the walking line clearly?
3. What were the weather conditions? Temperature, wind, precipitation?
4. What footwear was the subject wearing? Were heels over two inches removed?
5. Did the subject have any documented physical condition affecting balance?
6. Did the officer demonstrate the test before administering it?
7. Did the officer read the standardized instructions in full, in the correct order?
8. Did the HGN stimulus move at the NHTSA-specified speed and distance?
9. Were all three tests administered, or only some?
10. Is there dashcam footage that shows the actual administration and the environment?

These questions are not hypothetical. Every single one maps to a specific section of the NHTSA Standardized Field Sobriety Testing manual. Each one is either documented on the scene or not. If it's not documented, that itself is a question.

What a Blog Post Can't Do

This post gives you the framework. Three tests. Their NHTSA-published accuracy numbers. The conditions that invalidate them. The protocol deviations that break the validation.

Here's what the Field Sobriety Test Accuracy Review gives you that this blog post can't: your specific tests, your conditions, your state's exact protocols, mapped against NHTSA's own validation requirements. Not general theory. Your stop, your surface, your footwear, your physical condition, the specific tests the officer administered, and whether each one falls inside or outside the conditions under which NHTSA says the test actually works.

Not with rage. With a system. A forensic checklist built from your case facts and NHTSA's own manual.

It's $97. And it tells you exactly which questions to ask before you sit down with anyone about your case.

---

This article provides general information about NHTSA's published field sobriety test standards, not legal advice. Whether a specific test may be challenged in a specific case depends on the facts of that stop and the applicable state procedures.

Not sure where your case stands? Take the free Defense Milestone Score, 10 questions, instant results, no email required.