INTRODUCTION: Biological tests of drug use can be used to inform clinical and legal decisions and hold potential to provide evidence for epidemiological studies where self-reported behaviour may be unavailable or unreliable. We test whether hair can be considered as a reliable marker of cannabis exposure. METHODS: Hair samples were collected from 136 subjects who were self-reported heavy, light or non-users of cannabis and tested using GC-MS/MS. Sensitivity, specificity, positive predictive value and negative predictive value were calculated for five cannabinoids (tetrahydrocannabinol [THC], THC-OH, THC-COOH, cannabinol and cannabidiol). Samples also were segmented in 1 cm sections representing 1 month exposure and the correlation between amount of cannabinoid detected and self-reported cannabis consumption tested. RESULTS: All five cannabinoids were detected. Seventy-seven percent of heavy users, 39% of light users and 0% of non-users tested positive for THC. The sensitivity of detection of THC was 0.77 (0.56-0.91) comparing heavy cannabis smokers with light and non-users, whereas the sensitivity of other cannabinoids generally was considerably lower. The positive and negative predictive value of detection of THC were 0.57 (0.39-0.74) and 0.91 (0.82-0.97), respectively. A correlation of 0.52 (P < 0.001) was observed between self-reported monthly cannabis use and THC. DISCUSSION: Hair analysis can be used as a qualitative indicator of heavy (daily or near daily) cannabis consumption within the past 3 months. However, this approach is unable to reliably detect light cannabis consumption or determine the quantity of cannabis used by the individual. [Taylor M, Lees R, Henderson G, Lingford-Hughes A, Macleod J, Sullivan J, Hickman M. Comparison of cannabinoids in hair with self-reported cannabis consumption in heavy, light and non-cannabis users. Drug Alcohol Rev 2016;00:000-000].