Background and Aims Studies often rely upon self‐report and biological testing methods for measuring illicit drug use, although evidence for their agreement is limited to specific populations and self‐report instruments. We aimed to examine comprehensively the evidence for agreement between self‐reported and biologically measured illicit drug use among all major illicit drug classes, biological indicators, populations and settings. Methods We systematically searched peer‐reviewed databases (Medline, Embase and PsycINFO) and grey literature. Included studies reported 2 × 2 table counts or agreement estimates comparing self‐reported and biologically measured use published up to March 2022. With biological results considered to be the reference standard and use of random‐effect regression models, we evaluated pooled estimates for overall agreement (primary outcome), sensitivity, specificity, false omission rates (proportion reporting no use that test positive) and false discovery rates (proportion reporting use that test negative) by drug class, potential consequences attached to self‐report (i.e. work, legal or treatment impacts) and time‐frame of use. Heterogeneity was assessed by inspecting forest plots. Results From 7924 studies, we extracted data from 207 eligible studies. Overall agreement ranged from good to excellent (> 0.79). False omission rates were generally low, while false discovery rates varied by setting. Specificity was generally high but sensitivity varied by drug, sample type and setting. Self‐report in clinical trials and situations of no consequences was generally reliable. For urine, recent (i.e. past 1–4 days) self‐report produced lower sensitivity and false discovery rates than past month. Agreement was higher in studies that informed participants biological testing would occur (diagnostic odds ratio = 2.91, 95% confidence interval = 1.25–6.78). The main source of bias was biological assessments (51% studies). Conclusions While there are limitations associated with self‐report and biological testing to measure illicit drug use, overall agreement between the two methods is high, suggesting both provide good measures of illicit drug use. Recommended methods of biological testing are more likely to provide reliable measures of recent use if there are problems with self‐disclosure.