Phobia/panic disorder is common. It improves with exposure therapy, even when guided mainly by a computer self-help system such as FearFighter (FF), but such therapy must also demonstrate cost-effectiveness. This study compares the cost-effectiveness of FF with computed-aided relaxation and clinician-led exposure. Data were obtained on patients from a randomised controlled trial of FF. Economic analyses used pretreatment and 1-month follow-up self-ratings of the main problem and global phobia. Clinician costs were calculated using the number of therapist hours and the cost of FF. Incremental cost-effectiveness ratios were calculated and cost-effectiveness acceptability curves were produced. Data were available on 62 patients with main problem ratings and 60 with global phobia ratings. FF and clinician-led exposure were more effective than relaxation but more expensive. Compared with relaxation, producing an extra unit of improvement on the main problem scale cost 64 UK pounds with FF and 100 UK pounds with clinician-led exposure. FF appeared to be more cost-effective using the global phobia rating (112 UK pounds per extra unit of improvement vs. 128 UK pounds for clinician-led exposure). The cost-effectiveness of FF could be enhanced if users had less highly trained supporters. FF would be less cost-effective if face-to-face therapy was delivered by less qualified professionals. Caution is urged regarding these indicative findings given that these were secondary analyses.