Although motivational interviewing (MI) has been shown to be effective for a variety of issues, it is unclear how it works. As such, this experiment used a dismantling approach to examine hypothesized change mechanisms in MI. Seventy-two undergraduate participants who scored in the top 35th percentile on trait anger were randomly assigned to one of three conditions: full MI (FMI), spirit only MI (SOMI), or psychoeducation. They met individually with an experimenter for 30 to 45-minute sessions to discuss their anger. Sessions were recorded and coded by double-blind research assistants. In the FMI condition, the relational and technical elements of MI were both used to elicit change talk. In the SOMI condition, the supportive and relational elements of MI were emphasized. In the psychoeducation condition, focus was placed on teaching the components of anger episodes. Participants were then asked to launch a daily, online, five-minute deep breathing exercise during the following week. Results did not support a main effect on participant launches. Participants in both MI conditions, however, emitted more change talk than participants in the psychoeducation condition. Independent session ratings showed that despite the attempted elimination of technical elements in the SOMI condition, the FMI and SOMI conditions did not differ. These results, along with others, indicate it may be impossible to separate relational from technical elements in MI and, thus, to identify core mechanisms of change.