NHTSA Field Sobriety Test Clues Explained — What Officers Look For and How Accurate the Tests Really Are
When a police officer pulls someone over on suspicion of drunk driving, the roadside evaluation that follows is not a guessing game. It is a carefully developed, scientifically validated process — and the specific behaviors officers watch for, known as clues, are the backbone of that process. Understanding what those clues are, where they come from, and how reliably they help officers make the right call is something every driver deserves to know. This article breaks down the NHTSA field sobriety test clues, explains what each test involves, and shares the concrete numbers from a major real-world validation study conducted on the streets of San Diego.
Much of the data in this article comes from the NHTSA-commissioned field validation research by Jack Stuster and Marcelline Burns, which tested the accuracy of these tools against real breath and blood test results across nearly 300 traffic stops.
The Three Tests That Make Up the SFST Battery
The Standardized Field Sobriety Test (SFST) battery consists of three tests developed through NHTSA-sponsored research and first validated in 1981. Each test is designed to detect specific signs of alcohol impairment — not to simulate driving, but to give officers a reliable way to estimate whether a motorist’s blood alcohol concentration (BAC) is above the legal limit.
The three tests are:
- Horizontal Gaze Nystagmus (HGN) — an eye-tracking test that detects involuntary jerking of the eyes caused by alcohol
- Walk and Turn (WAT) — a divided-attention test requiring the subject to walk heel-to-toe along a line and return
- One Leg Stand (OLS) — a balance and counting test in which the subject holds one foot off the ground for 30 seconds
Each test has a defined set of clues. If a subject shows enough clues, the officer interprets that as an indication the motorist’s BAC is at or above the legal threshold.
NHTSA Field Sobriety Test Clues: What Officers Are Watching For
Horizontal Gaze Nystagmus — 6 Possible Clues
The HGN test tracks how a person’s eyes move when following a slowly moving object. Alcohol interferes with the brain’s control of eye movement, producing an involuntary jerking called nystagmus. Officers check three indicators in each eye, for a maximum of six clues total:
- Lack of smooth pursuit (the eye jerks rather than following smoothly)
- Distinct and sustained nystagmus at maximum deviation (jerking when the eye is turned as far as it can go)
- Onset of nystagmus before the eye reaches 45 degrees from center
Under the modified scoring used in the San Diego validation study, four or more HGN clues indicated a BAC of 0.08 percent or higher. Two or more clues suggested a BAC above zero but below 0.08 percent.
Walk and Turn — 8 Possible Clues
The walk and turn clues NHTSA officers look for are designed to expose the kind of difficulty an impaired person has when trying to follow instructions and coordinate movement at the same time. The test requires nine heel-to-toe steps along a straight line, a specific turn, and nine steps back. Officers watch for eight distinct behaviors:
- Cannot keep balance while listening to instructions
- Starts walking before the instructions are finished
- Stops while walking to regain balance
- Does not touch heel-to-toe on at least one step
- Steps off the line
- Uses arms to balance (raising them more than six inches from the sides)
- Makes an improper turn
- Takes an incorrect number of steps
Two or more of these clues is the threshold for estimating a BAC at or above 0.08 percent. The test is a “divided attention” exercise — it demands that a subject listen, remember, and move precisely all at once, which alcohol makes significantly harder.
One Leg Stand — 4 Possible Clues
In the one-leg stand test, the subject holds one foot about six inches off the ground and counts aloud (one-thousand-one, one-thousand-two, and so on) for 30 seconds. Officers watch for four signs of impairment:
- Swaying while balancing
- Using arms to balance
- Hopping to maintain balance
- Putting the raised foot down before the 30 seconds are up
Two or more clues from this test also indicates a likely BAC at or above 0.08 percent.
Why a New Study Was Needed in the 1990s
The original SFST battery was validated in 1981 — but at that time, it was calibrated to a 0.10 percent BAC threshold. As states across the country began lowering their legal limit to 0.08 percent (and adopting zero-tolerance laws for drivers under 21), the question became pressing: do these same tests still work reliably at these lower levels?
That question prompted NHTSA to fund the San Diego study, which set out to see whether the battery — with slightly adjusted scoring — could accurately help officers make arrest decisions at the 0.08 percent level, and even at 0.04 percent.
How the San Diego Study Was Conducted
Seven experienced officers from the San Diego Police Department’s dedicated alcohol enforcement unit participated. All had received NHTSA-approved SFST training and had at least three years in the Traffic Division before joining the special unit. Over the course of the study — from May 23 to November 9, 1996 — officers administered the SFST battery to motorists during routine patrols, following a strict protocol.
Crucially, every officer was required to estimate the driver’s BAC based solely on SFST results before any breath test was given. This prevented test results from influencing the estimates. After the estimate was recorded, a portable evidentiary breath test confirmed (or contradicted) the officer’s assessment.
A total of 298 data collection forms were completed. One case was removed because the motorist refused all BAC testing, leaving 297 valid cases for analysis. Of those drivers:
- 73% were arrested for DWI
- 22% received warnings
- 5% were cited for other vehicle violations
The Accuracy Numbers: How Well Do the Tests Actually Work?
Overall Estimation Accuracy
Across all 297 motorists tested, officers’ mean estimated BAC was 0.117 percent. The mean measured BAC was 0.122 percent. That gap — just 0.005 percent — is smaller than the margin of error on most evidentiary testing devices and is considered operationally irrelevant. Officers were, on average, remarkably close to the truth.
Decision Accuracy at the 0.08 Percent Level
When it came to the key legal threshold of 0.08 percent BAC, here is how the results broke down:
| Decision Type | Correct Decisions | Accuracy Rate |
|---|---|---|
| Overall decisions (all 297 cases) | 269 out of 297 | 91% |
| Decisions to arrest (BAC estimated ≥ 0.08%) | 210 out of 234 | 90% |
| Decisions not to arrest (BAC estimated < 0.08%) | 59 out of 63 | 94% |
When 10 borderline false positive cases are taken into account — including one driver who was a juvenile (where any detectable BAC is a legal violation) and nine drivers with BACs between 0.07 and 0.08 (within testing device margins) — the overall accuracy climbs to 94 percent.
There were 24 false positives (officers estimated BAC ≥ 0.08% but measured BAC was below) and only 4 false negatives (officers estimated below 0.08% but measured BAC was above). Six of the false positives had actual BACs of exactly 0.07 percent — just a single point below the threshold.
How Each Individual Test Performed
| Test | Overall Accuracy | “Arrest” Decision Accuracy | “Release” Decision Accuracy |
|---|---|---|---|
| Horizontal Gaze Nystagmus (HGN) | 88% | 87% | 93% |
| One Leg Stand (OLS) | 83% | 86% | 73% |
| Walk and Turn (WAT) | 79% | 82% | 69% |
Correlation with Actual BAC
Statistical correlation analysis — measuring how closely each test’s scoring tracked actual measured BAC — produced the following results:
| Test(s) | Correlation with Measured BAC (r) |
|---|---|
| All Three Tests Combined | 0.69 |
| Horizontal Gaze Nystagmus | 0.65 |
| Walk and Turn | 0.61 |
| One Leg Stand | 0.45 |
For field tests administered by officers at the roadside, correlations in the 0.65–0.70 range are considered very high. The data make clear that using all three tests together consistently outperforms any single test on its own.
The HGN Test Is the Strongest Predictor — Here Is Why That Matters
HGN consistently emerged as the most reliable individual predictor of BAC, both in terms of correlation (r=0.65) and overall decision accuracy (88%). One key reason is that nystagmus is an involuntary physiological response — a driver cannot consciously suppress or fake it. Experienced drinkers can sometimes mask obvious behavioral impairment, but they cannot prevent alcohol from triggering the eye-movement patterns that HGN detects.
Some courts have been skeptical of HGN because it does not look like a driving-related test. The research addresses this directly: the purpose of the SFST battery is not to simulate driving but to reliably estimate BAC. Since DWI laws are written around BAC thresholds — not around whether a specific driver appears impaired — a test that accurately predicts BAC is exactly what the law calls for.
In 162 of the 188 cases where the driver’s measured BAC was at or above 0.08 percent, all three component tests agreed with each other. When discrepancies occurred, HGN was the source of the disagreement far less often than either the Walk and Turn or the One Leg Stand — it was involved in only 4 disagreements, compared to 16 for Walk and Turn and 15 for One Leg Stand.
What About Lower BAC Levels — Does It Work Below 0.08 Percent?
A secondary goal of the study was to see whether the SFST battery could also help officers identify drivers with BACs between 0.04 and 0.08 percent — relevant in states with tiered DWI statutes or zero-tolerance laws for young drivers. Officers used two HGN clues (rather than four) as the threshold for estimating a BAC above 0.04 percent.
Among the 83 cases where this threshold was relevant, officers’ decisions to arrest were accurate 94 percent of the time. Overall accuracy in distinguishing above-0.04 from below-0.04 cases was 80 percent. The smaller dataset at this lower threshold means slightly less certainty in the results, but they are consistent with findings from a similar Colorado study where 93 percent of arrest decisions at the 0.05 percent level were correct.
Officers and Prosecutors: Both Accepted the Results
Beyond the statistics, the study assessed how officers and legal professionals actually felt about using the SFST battery at the lower BAC thresholds. Every participating officer reported full confidence in the tests for field use. They noted the battery could be administered under virtually all reasonable weather and road conditions, and that their daily experience using the tests reinforced their reliability.
Representatives from the San Diego City Attorney’s Office confirmed that none of the 298 DWI cases from the study period was negatively affected by using only the three NHTSA SFSTs. Defense attorneys had been unsuccessful challenging the NHTSA battery in court — in contrast to other non-standardized tests used locally, which defense counsel had successfully challenged due to lack of scientific validation. Prosecutors suggested that all agencies should limit DWI evaluations to the standardized NHTSA battery for exactly this reason.
Conclusion: A Validated System Built on Real Evidence
The NHTSA field sobriety test clues are not arbitrary — they are the product of decades of research, refined through laboratory studies, field trials, and legal scrutiny. The San Diego study demonstrated that a trained officer using the three-test SFST battery can accurately identify whether a motorist is above or below the 0.08 percent BAC threshold 91 to 94 percent of the time.
Understanding the walk and turn clues NHTSA officers use — along with the other SFST components — helps demystify a process that can feel opaque and high-stakes for anyone who encounters it at the roadside. These tests are not perfect, but the data show they are a sound, scientifically defensible foundation for DWI enforcement, especially when all three components are used together.
If you’ve ever wondered why a particular behavior triggers an officer’s attention during a traffic stop, the answer lies in years of research showing that those precise clues are meaningfully linked to alcohol in the system — not instinct, not arbitrary judgment.