Befriending (BF) has received attention as a beneficial part of mental health care. For example, when compared with cognitive-behavioural therapy (CBT) for schizophrenia, BF was found to be as effective as CBT in terms of initial symptom improvement (Sensky et al., 2000). Why was the BF apparently so effective? Was the BF really a discrete and powerful intervention? Or was it simply an unrecognized aspect of regular CBT? We addressed these questions by first observing BF's convergence with “social support” (a plausibly discrete treatment) and, second, by examining its divergence from CBT, drawing on archival data. For the convergence prediction we correlated the speech content of therapists' in 10 BF sessions from the Sensky et al. study with previously published social support data from stylists working with people with severe mental health problems in a psychiatric hospital (Milne and Netherwood, 1997). For the second prediction the same 10 BF sessions were compared with a sample of 10 CBT sessions, also from the Sensky et al. study. The results indicated that BF was indeed significantly correlated (converged) with social support (r=0.7; p<.05), and also that it did differ (diverged) significantly from CBT (p<.05). As a treatment fidelity check, therefore, these findings indicate tentatively that the two treatments in the Sensky et al. (2000) study were appropriately implemented. As an analysis of effective interventions for schizophrenia, it cautiously suggests that “social support” merits serious attention, and is perhaps no mere placebo.