AIM: To assess the properties of the Cigarette Dependence Scale (CDS-12) in various samples of daily smokers and to provide reference scores in a general population sample. METHODS: Surveys in 4 samples of daily cigarette smokers: psychiatric out-patients in Geneva, Switzerland (n=226), clients of smoking cessation clinics in France (n=370), visitors of a French-language smoking cessation website (n=13,697) and a representative sample of the general population of Geneva (n=292). RESULTS: In all 4 samples, Cronbach's alpha coefficients were >0.87 and factor analyses indicated that CDS-12 was unidimensional. CDS-12 was slightly skewed towards higher values, and it was associated with expired carbon monoxide, but this association was not strong (9% of variance explained, p<0.001). CDS-12 scores were highest in clients of smoking cessation clinics (mean=47.7, SD=10.2), followed by psychiatric patients (mean=44.4, SD=8.4), visitors of smoking cessation websites (mean=43.3, SD=11.6) and the general population sample (mean=36.9, SD=12.3). Except for tolerance, each element in the DSM-IV and ICD-10 definitions of dependence is reflected by at least one item in CDS-12, even though the match with these definitions is sometimes indirect. CONCLUSIONS: This paper presents reference scores and validity and reliability tests for CDS-12 in a diversity of samples of daily smokers. This information should be useful to clinicians and researchers.