A scoping study, the National Human Exposure Assessment Survey (NHEXAS) was conducted in EPA Region V from July 1995 to May 1997. This probability-based population study provided an opportunity to examine the mercury levels in 182 participants who provided hair samples. A sensitive analytical procedure based on atomic fluorescence spectrometry was developed and evaluated for the analysis of Hg in approximately 5 mg of human hair. The correlation coefficient (r), the precision, and bias were 0.9983, 1.6%, and 8%, respectively, for standard curves in the hair matrix. The method detection limit (MDL), recovery of Hg in a certified sample (NIES-13), precision (% RSD) for duplicate extract analysis, and precision for duplicate sample analysis averaged 12 ppb (range 4 to 22 ppb), 100 3% (N =27), 4.6 2.8 (N =18), and 12.5 7.4 (N =17), respectively, over the 7 to 8 months of sample analysis. The low MDL yielded 95% of the samples with measurable values, permitting the entire distribution of Hg levels to be characterized. Comparison of annualized Hg distribution in hair with and without background correction revealed a negligible bias on the distribution (1.47% at the 90th percentile). Also, a comparison of the unweighted and nonannualized weighted Hg levels throughout the percentile distribution indicated a small deviation in the upper tail (95th percentile) and is attributable to the small sample size (N =182). The mean, median, and maximum of the annualized Hg levels in hair were 287, 204, and 3505 ppb, respectively. The 75th percentiles were 335 and 368 ppb for the weighted annualized and unweighted distributions, respectively. The percent of individuals in three age categories (0 ±24, 25 ± 49, and 50 years and older) who exceeded the 75th percentile showed a linear increase with age. Males (N =81) had 10% and 20% lower mean levels than females (N =101) for unweighted and annualized weighted Hg data, respectively. The application of this methodology for characterizing hair Hg levels in fish-eating populations is discussed.