Add-back hormone replacement therapy has been shown to alleviate some of the hypo-oestrogenic side effects associated with gonadotrophin-releasing hormone agonists, including demineralisation of bone. Studies on patients with uterine fibroids have shown that concomitant add-back therapy reduced the efficacy of these agents, but that deferred administration was less detrimental. This trial set out to investigate if deferred add-back therapy could offer any advantages to patients with endometriosis compared with immediate therapy. Zoladex [goserelin acetate (3.6 mg every 4 weeks)] was given for 24 weeks either with placebo, with medrogestone (10 mg/day) for 24 weeks (immediate add-back therapy), or with placebo for 12 weeks followed by medrogestone (10 mg/day) for 12 weeks (deferred add-back therapy) to 123 patients. The number of responders measured using the Revised American Fertility Society score (decrease in this score of > or = 50%) was greatest in the immediate add-back therapy group, although there were no significant differences between groups. All three treatment groups showed significant decreases in bone mineral density compared with baseline but smaller losses were generally observed in the add-back groups. A significantly smaller number of patients in the immediate add-back group reported hot flushes during the first 12 weeks of treatment compared with the deferred add-back group. In conclusion, it appears that there is no extra advantage to patients with endometriosis being treated with goserelin in delaying the start of add-back therapy.