Few long-term treatment regimens for severe acne vulgaris have been investigated in clinical trials. Data were combined from two consecutive, randomized, double-blind, controlled studies to evaluate the efficacy, safety and subject satisfaction of four nine-month regimens in severe acne vulgaris treatment. Subjects were first randomized to receive doxycycline (DCN) and adapalene 0.1% - benzoyl peroxide 2.5% (A/BPO) or vehicle once daily for 12 weeks. Subjects who had at least 50% global improvement were subsequently randomized to receive A/BPO or its vehicle once daily for 24 weeks. Over nine months, there were four regimens: A/BPO and DCN followed by A/BPO, vehicle and DCN followed by A/BPO, A/BPO and DCN followed by vehicle, and vehicle and DCN followed by vehicle. Among the four regimens, A/BPO and DCN followed by A/BPO led to the highest percentage of subjects rated "clear" or "almost clear" (50.0% vs. 40.4%, 26.2% and 25.0%, respectively), biggest reduction in total lesion counts (76% vs. 70%, 51% and 47%, respectively) and greatest subject satisfaction (85.0% vs. 75.5%, 63.3% and 52.4%, respectively) at week 36. It provided a faster onset of action compared to groups started with vehicle and DCN (P<.05 at week 2). Subjects receiving A/BPO and DCN followed by vehicle experienced deterioration once the active treatment was discontinued. All regimens were safe and well-tolerated. In conclusion, efficacious initial therapy and long-term treatment are both important. An initial combination therapy with adapalene-BPO and DCN followed by longer-term adapalene-BPO treatment is an efficacious and satisfactory new regimen for severe acne subjects.