Estimates of the prevalence of alcohol dependence among Americans approach 14% (Read, Kahler, & Stevenson, 2001). Alcohol dependence was once considered among the most recalcitrant of problem behaviors, with only 20% to 30% attaining sustained abstinence (Hunt Barnett & Branch 1971). Although current definitions of treatment success now consider lapses and recycling (e.g., DiClemente, 2003), sustained abstinence remains the gold standard and is achieved in up to 60% of people in efficacy trials of current psychosocial treatments (e.g., Project MATCH). This article describes our efforts to develop the next-generation CBT treatment manuals for patients in early abstinence from alcohol, Broad Spectrum Treatment (BST). BST attempts to simultaneously address two seemingly incompatible treatment research goals. First, BST is a flexible but manual-guided treatment that can be standardized and used in the field with the broad spectrum of alcohol abusers. Second, BST seeks to maximize treatment effectiveness by tailoring the specific treatment package to the individual patient's needs and capacities. Use of explicitly defined a priori decision trees is the vehicle through which these goals can be accomplished. This article describes the manual in its current form, and discusses the manner in which we are presently testing the efficacy of the manual as it stands.