Objectives: This study focuses on the predictive and comparative significance of ethyl glucuronide measured in head hair (hEtG) for estimating risks associated with alcohol-impaired driving offenders. Earlier work compared different alcohol biomarkers for estimating rates of failed blood alcohol concentration (BAC) tests logged during 8 months of interlock participation. These analyses evaluate the comparative performance of several alcohol markers including hEtG and other markers, past driver records, and psychometric assessment predictors for the detection of 4 criteria: new driving under the influence (DUI) recidivism, alcohol dependence, and interlock record variables including fail rates and maximal interlock BACs logged. Methods: Drivers charged with alcohol impairment (DUI) in Alberta, Canada (n = 534; 64% first offenders, 36% multiple offenders) installed ignition interlock devices and consented to participate in research to evaluate blood-, hair-, and urine-derived alcohol biomarkers; sit for interviews; take psychometric assessments; and permit analyses of driving records and interlock log files. Subject variables included demographics, alcohol dependence at program entry, preprogram prior DUI convictions, postenrollment new DUI convictions, self-reported drinking assessments, morning and overall rates of failed interlock BAC tests, and maximal interlock BAC readings. Recidivism, dependence, high BAC, and combined fail rates were set as criteria; other variables were set as predictors. Area under the receiver operating characteristics (ROC) curve (A′) estimates of sensitivity and specificity were calculated. Additional analyses were conducted on baseline hEtG levels. Driver performance and drinking indicators were evaluated against the standard hEtG cutoff for excessive drinking at (30 pg/mg) and a higher criterion of 50 pg/mg. HEtG splits were evaluated with the Mann-Whitney rank statistic. Results: HEtG emerged as a top overall predictor for discriminating new recidivism events that occur after interlock installation, for entry alcohol dependence, and for the highest interlock BACs recorded. Together, hEtG and phosphatidylethanol (PEth) were the top predictors of all criterion measures. By contrast, the hair-derived alcohol biomarkers hEtG and hFAEE (fatty acid ethyl esters) were poorer than other alcohol biomarkers as detectors of interlock BAC test fail rates. Conclusions: This study showed that hEtG, an objective alternative to often unreliable self-reported past representation of drinking levels, yields crucial insight into driver alcohol-related risks early in an interlock program and is a top predictor of new recidivist events. Together with PEth, these markers would be excellent anchors in a panel for detecting alcohol consumption.