The definition of an empirically supported treatment (EST) arguably embodies 2 untested assumptions: (a) that different manualized renditions of the same therapy are functionally equivalent and (b) that therapies can be reliably applied independently of therapist, setting, and format. These assumptions were tested as applied to cognitive therapy (CT), using process data from a large multisite study (N = 235) that included 3 cognitive and 6 alternative therapies. Although the non-CTs were more variable than the CTs on 2 of 4 dimensions studied (directiveness and emotional arousal), there was considerable variation among the 3 CTs, even when implemented in the current context of rigorous training, manualization, and adherence checks. Results are discussed as related to the assumptions underlying EST criteria.