OBJECTIVE: To determine whether participants with severe head injury (SHI) allocated to a brief compassion focused imagery (CFI) intervention show greater change in compassion than those exposed to relaxation imagery (RI). METHOD: Participants were exposed to a preparatory video to promote engagement and then randomly allocated to intervention. Pre- and post-preparatory measures were Motivation for Intervention and Fears of Compassion Scales, State-Trait Anxiety Inventory (STAI) and PANAS. Pre- and post-intervention self-report measures were the Empathy Quotient, Self-Compassion Scale, STAI and Relaxation Scale. Heart rate variability (HRV) was monitored throughout. RESULTS: Motivation for therapy increased after the preparatory video (z = 3.44, p = 0.001). Across the intervention, group differences were not found on self-report measures or HRV changes. When CFI and RI groups were pooled, improvement in relaxation (r = .41, p < 0.01) and state anxiety (r = .29, p < 0.05) were found across the intervention; these outcomes were not associated with changes in self-compassion or HRV. CONCLUSION: Brief CFI, a central aspect of compassion focused therapy, did not produce a reliable change in people with SHI. Enhanced motivation for psychological therapy after a brief preparatory video is relevant and underlines the need to understand mechanisms of action rather than the pursuing whole protocol approaches to therapy.