In a randomized controlled trial of breast cancer patients awaiting surgery, 51 client-centered psychotherapeutic interventions and 56 "chats" were analyzed using Stiles' Verbal Response Modes (VRMs). The 30-minute interventions and chats were carried out the night before surgery by a consultant surgeon trained in client-centered counseling techniques. Significant differences in therapist and client VRMs were found between interventions and chats. Therapist reflection and acknowledgment were the predominant VRMs in the interventions. Significantly more therapist reflection and acknowledgment occurred in the interventions than in the chats, and were associated with significantly more client disclosure. However, intervention and chat patients were indistinguishable on most outcome measures. Some chats bore a resemblance to reminiscence therapy with the elderly. In a content analysis carried out on the chats, the most common topics were holidays, followed by family. Neither VRM disclosure, personal topics discussed, nor "significant" disclosures in the chats predicted better therapeutic out-