A potential problem in the clinical use of hair assays has been that drug concentration in hair may have a racial bias. This has been linked primarily to cocaine identification and recovery from hair. This hypothesis has appeared in several studies but there are major flaws that exist in the relationship between race and cocaine values in hair assays. These flaws consist of theoretical constructs of “race” used by the proponents of bias theory. A handful of researchers interested in hair analysis have resurrected an archaic typology in the analysis of a racial bias effect in hair testing. There are inadequacies of both the methodology and the data utilized to demonstrate this bias effect. There is also the failure of these biases to manifest themselves in clinical programs that involve the assessment of large populations of persons on whom detailed drug-taking histories and behaviors are known. A sample group was studied for contrasts between African-American and white subjects. One observation was that, with regard to hair color, there was no clear link between specific hair color and hair analysis outcome for cocaine. Overall, hair color appeared to make only a very small contribution to outcome variance. With regard to race, both African Americans and whites were most likely to test negative for cocaine and to self-report no cocaine use. African Americans were more likely to test cocaine positive by any bioassay when compared to whites. The ratios between self-reports and assays were roughly concordant across race. In some studies, there is a significant difference between African-American and white arrestees in mean value for the cocaine hair assay for the sample as a whole. These findings continue to generally support the efficacy of both hair and urine assays as good identifiers of cocaine use. It appears that hair color variance is probably a factor in cocaine retention but not a significant one. Racial bias as a clinically significant factor in the interpretation of hair assay values for cocaine use does not represent a serious impediment in most applications. 16 tables, 48 references