Objective: Clinicians commonly "drift" away from using proven therapeutic techniques. This study examined the degree to which such drift occurs among cognitive behavioral therapy (CBT) clinicians working with a specific clinical population--adults with eating disorders. Method: The study used a correlational design. The participants were 80 qualified clinicians (69 women, 11 men; mean age = 39.2 years, range = 23-62 years) who routinely offered what they described as CBT to adults with eating disorders. Each clinician detailed whether and how often he or she used different cognitive behavioral techniques when delivering CBT to such patients, and each completed the anxiety scale of the Brief Symptom Inventory (Derogatis, 1983). Results: Implementation of specific CBT techniques was far lower than protocols would suggest, particularly for clinicians who were anxious, older, or more experienced in working with the eating disorders (p less than 0.05, in all cases). The use of treatment manuals was associated with greater use of recommended CBT techniques (p less than 0.05, in all cases). Cluster analysis showed that clinicians fell into three types--behavior, motivation, and mindfulness oriented. Conclusions: These findings need to be extended to other therapies and other disorders, but they indicate the need for stronger training and closer supervision if clinicians are to give patients the best chance of recovery. They demonstrate that clinicians' use of the label "CBT" is not a reliable indicator of the therapy that is being offered. (Contains 3 tables.)