OBJECTIVE: To evaluate who benefits most from exposure-focused cognitive behavior therapy (CBT) for pathological health anxiety. METHODS: This was a secondary study based on a randomized controlled trial of Internet-delivered versus face-to-face CBT (N = 204) delivered in primary care for participants with pathological health anxiety and a primary diagnosis of either somatic symptom disorder or illness anxiety disorder. Linear mixed effects regression models were fitted on 13 weekly 18-item Health Anxiety Inventory assessments over the treatment phase. The time×variable coefficient was tested as indicative of a predictive effect, regardless of treatment format. The treatment format (Internet versus face-to-face) was then added to the model, and the time×format×variable coefficient was tested as indicative of moderation of the between-format effect. The following pre-treatment variables were evaluated: age, gender, education, health anxiety, subtypes of health anxiety behaviors, anxiety sensitivity, intolerance of uncertainty, depression symptoms, experience of psychotherapy, noncommunicable disease, treatment preference, and path of referral. RESULTS: Regardless of delivery format, a larger reduction in health anxiety was predicted by higher pre-treatment health anxiety, bodily preoccupation/checking, prevention/planning, anxiety sensitivity, and depression symptoms. Most effects were small. None of the candidate variables were significant moderators of the relative effect of Internet-delivered and face-to-face CBT. CONCLUSION: This study replicates that higher pre-treatment health anxiety and anxiety sensitivity are predictive of a larger reduction in health anxiety regardless of CBT format. So far, no robust moderators of the relative effect of Internet-delivered versus face-to-face CBT for health anxiety have been identified.