Background: Rates of treatment seeking for cannabis are increasing, and relapse is common. Management of cannabis withdrawal is an important intervention point. No psychometrically sound measure for cannabis withdrawal exists, and as a result treatment developments cannot be optimally targeted. The aim is to develop and test the psychometrics of the Cannabis Withdrawal Scale and use it to explore predictors of cannabis withdrawal. Methods: A volunteer sample of 49 dependent cannabis users provided daily scores on the Cannabis Withdrawal Scale during a baseline week and 2 weeks of abstinence. Results: Internal reliability (Cronbach's alpha=0.91), test-retest stability (average intra-class correlation=0.95) and content validity analysis show that the Cannabis Withdrawal Scale has excellent psychometric properties. Nightmares and/or strange dreams was the most valid item (Wald χ 2=105.6, P<0.0001), but caused relatively little associated distress (Wald χ 2=25.11, P=0.03). Angry outbursts were considered intense (Wald χ 2=73.69, P<0.0001) and caused much associated distress (Wald χ 2=45.54, P<0.0001). Trouble getting to sleep was also an intense withdrawal symptom (Wald χ 2=42.31, P<0.0001) and caused significant associated distress (Wald χ 2=47.76, P<0.0001). Scores on the Severity of Dependence Scale predicted cannabis withdrawal. Conclusions: The Cannabis Withdrawal Scale can be used as a diagnostic instrument in clinical and research settings where regular monitoring of withdrawal symptoms is required.