Serum from patients with aplastic anaemia contains two distinct stimulatory activities on haemopoiesis in culture. The first is a highly unstable enhancing activity, which mainly stimulates colony formation from BFU-E and macrophage precursors, and only acts when added directly to target bone marrow cultures. It is destroyed by Sephadex G-150 chromatography, and thus differs from colony stimulating activity (CSA) and burst promoting activity (BPA). Its mode of action is unknown. It was elevated in 70/97 patients with severe aplastic anaemia (SAA). 71 of these 97 patients were treated with high dose immunosuppression. 55/71 who achieved self-sustaining haemopoiesis had higher serum stimulating activity on BFU-E than 16/71 who never achieved remission (P = 0.0004). It was predictive of response as an all or none phenomenon, independent of the time required for recovery. It was, however, unsatisfactory as a single prognostic test. Bone marrow reconstitution also occurred in 4/71 patients whose serum inhibited BFU-E in direct culture. The second stimulator acts via enhancement of CSA- and BPA-release by accessory cells and is therefore termed 'releaser' activity. It was elevated in 27 of 51 aplastic anaemia patients and did not correlate with direct stimulatory activity. High 'releaser' activity was not predictive of response in 42 patients treated with high dose immunosuppression. However, the ability of patient cells to respond to autologous 'releaser' activity was a positive risk factor. Patients whose cells released an excess of CSA in the presence of autologous serum had a significantly higher chance of autologous recovery within 3 months than patients who produced little or no CSA in the presence of normal or excess releaser activity (P less than 0.0001). A scoring system which includes these two good risk factors is proposed for estimation of a patient's probability to recover autologous bone marrow reconstitution.