Aims: Our objective was to identify client characteristics and other factors associated with pre-treatment drop-out by people with marijuana dependence. Design and participants: Data from the Marijuana Treatment Project's screening assessment were used to examine correlates of pre-treatment drop-out. Information from all eligible study participants (n = 813) (i.e. those who were interested in receiving treatment for their marijuana dependence and were determined to be eligible for the randomized treatment efficacy trial) was used to examine differences between the 450 participants who initiated treatment (by enrolling in the trial) and the 363 individuals who declined enrollment. Setting: The study was conducted at three community-based outpatient treatment facilities in Farmington, CT, Seattle, WA and Miami, FL. Measurements: The information gathered in the screening interview included demographic characteristics, residential stability variables, employment and education history and referral source. Substance use variables included the number of days and the number of times per day marijuana was used, self-perceived dependence on marijuana, alcohol or other drugs, other drug use history and current treatment (i.e. substance abuse, medical, psychiatric) situation. Findings: Stepwise logistic regression was conducted to confirm variables associated with treatment initiation in bivariate analyses. Pre-treatment drop-out was associated with being younger, unmarried, unemployed, less educated and Asian American or Native American. It was also associated with self-perceived dependence on marijuana and use of other drugs. Conclusions: By recognizing demographic and substance use factors that may serve as barriers for individuals accessing treatment for marijuana dependence, clinicians may target clients with these characteristics proactively to encourage treatment initiation and subsequent attendance.