Patients with anorexia, bulimia, or eating disorder not otherwise specified (NOS) represent a psychiatrically and medically complex population. These disorders are characterized by chronicity, psychiatric comorbidities, and potentially serious medical and dental consequences carry the highest mortality rates of any psychiatric illness. Psychotherapy, particularly cognitive behavioral therapy (CBT), is the cornerstone of evidence based treatment. Variation is defined as the way the performance of a treatment process changes over time. While some random variation is normal, too much variation may result in deviation from fundamental CBT principles and may be less effective. Reducing non-random variation in group CBT delivery thus is an important part of maintaining fidelity to CBT principles and improving patient outcomes. The purpose of this project was to conceptualize and reduce causes of unwanted variability in the delivery of a twice weekly cognitive behavioral group therapy intervention, a core component of the partial hospitalization and intensive outpatient eating disorders programs at a Portland hospital.