Field Sobriety Test Accuracy: What the Numbers Really Show

NHTSA's training materials report about 91% accuracy for the combined standardized field sobriety test battery when detecting drivers at or above 0.08 BAC in one validation study, but that figure describes a screening result under study conditions, not a chemical measurement or a guarantee about any driver's case (NHTSA's 2023 SFST participant manual). That distinction matters: roadside tests record behavior that may support an inference of impairment, but they don't measure how much alcohol is in someone's blood. For a Connecticut DUI case, the practical question is not only how a test is scored. It's whether the reported clues came from alcohol-related impairment, another cause, or the way the test was administered. The video, medical history, scene conditions, and officer's notes can help test that conclusion. What a Field Sobriety Test Can and Cannot Prove A field sobriety test is a roadside observation of behavior. It produces scored clues about eye movement, balance, and divided attention, which is a different kind of evidence from a chemical reading of blood alcohol concentration. Officers commonly give the standardized tests in this order: horizontal gaze nystagmus (HGN), the walk-and-turn (WAT), and the one-leg stand (OLS). HGN involves watching the eyes as a pen or light moves across the person's field of vision. The WAT asks a person to take nine heel-to-toe steps, turn, and take nine steps back. The OLS asks the person to raise one foot and hold it up while counting. A smoke detector signals a possible fire, but it doesn't establish the cause or extent of the fire. The battery works the same way: HGN observes involuntary eye movement, while the WAT and OLS assess balance, coordination, divided attention, and the ability to follow instructions. The standardized battery grew out of research commissioned by the National Highway Traffic Safety Administration (NHTSA) in 1977 and was validated in controlled studies around the 0.10% BAC threshold in 1981. In the original research, the individual tests classified subjects with different accuracy figures: HGN 77%, WAT 68%, and OLS 65% (NHTSA full manual). Those rates describe classification within a study population; they say nothing about whether a particular driver was impaired. The manual also describes the original work's limited number of experienced officers and its laboratory and field settings, factors that make training and administration important to interpretation. The central limit: A roadside result is evidence to weigh, not a verdict or a substitute for chemical testing. Because the tests rely on observable behavior, each reliability concern points to a specific record that can confirm or rule it out. Medical conditions, physical limitations, and age appear in medical history. Footwear, weather, and road surface appear in scene photos or video. Whether instructions were demonstrated correctly and clues scored by the manual's criteria appears in the officer's notes and any recording of the stop. Drivers seeking a separate overview of roadside considerations can consult sobriety test guidance for NC drivers, while recognizing that procedures and legal consequences depend on jurisdiction. Accuracy Figures From the NHTSA Validation Studies Measure Reported accuracy What it classifies Horizontal gaze nystagmus (HGN) alone 77% (NHTSA full manual) Subjects at or above the alcohol threshold in original research Walk-and-turn (WAT) alone 68% (NHTSA full manual) Subjects at or above the alcohol threshold in original research One-leg stand (OLS) alone 65% (NHTSA full manual) Subjects at or above the alcohol threshold in original research Three-test battery, detecting drivers at 0.08 BAC or above About 91% (NHTSA 2023 participant manual) Detection in one validation study cited in NHTSA training material Three-test battery, Colorado validation (1995) 86% (NHTSA refresher participant manual) Correct arrest-or-release decisions Three-test battery, San Diego validation (1998) 95% (NHTSA 2023 participant manual) Correct arrest decisions under specified study conditions Florida (1997) validation No figure stated here The available NHTSA summary identifies the study but doesn't give its individual figure The first three figures describe individual-test classification in the original research: HGN scored 77%, WAT 68%, and OLS 65%. The combined-battery results address a different question. NHTSA reports about 91% detection at or above 0.08 BAC in one validation study, 86% correct arrest-or-release decisions in the 1995 Colorado validation, and 95% correct arrest decisions in the 1998 San Diego validation. These are study-specific classification or decision rates, not measurements of a driver's BAC. The single-test figures cannot be treated as the combined battery's accuracy, and neither set establishes that a particular person was impaired. Nor do these figures describe chemical-test accuracy. A population-level rate cannot show how often a specific stop produced a false positive. That depends on the tested population, the procedures used, and the evidence available in that case. The relevant comparison is between the study's measure and the actual record: what the officer administered and recorded, and what the video and other evidence show. Factors That Lower Field Sobriety Test Reliability A driver with a bad knee, heeled boots, and a gravel shoulder may struggle with the walk-and-turn even when alcohol played no part in the stumble. If the report labels that performance a sign of impairment, the record has to show what else could have produced each clue, and whether the officer accounted for it. Factor How it can distort the result Record that confirms or rules it out Age Balance and reaction may be affected, producing sway without intoxication Age in the report and medical history noting age-related balance limits Body weight and build Changes balance demands, especially on the one-leg stand Officer's notes on height and weight, booking records Injury or joint problems Knee, ankle, hip, or back limits leg-raising and turning Medical records, prior injury history, gait visible on video Inner ear or vision conditions Can cause sway or eye movement resembling HGN clues Vestibular or eye-condition diagnosis, corrective lenses noted Prescription medication or cannabis May cause sway or eye effects that roadside clues alone cannot trace to a substance Medication list, prescription records, timing of use Footwear Heels, boots, or sandals can reduce stability during the walk-and-turn Shoes described in the report and