We have compared three methods of preventing hypotension during subarachnoid anaesthesia. We attempted to maintain systolic arterial pressure (SAP) greater than 75% of baseline by use of i.v. fluids (preloading with normal saline 16ml kg(-1) and, if necessary, three subsequent boluses of 2.5 ml kg(-1)), an infusion of metaraminol titrated as necessary between 0 and 5 mg h(-1) and an infusion of ephedrine titrated as necessary between 0 and 120 mg h(-1). SAP and mean arterial pressure (MAP) were measured by automated oscillotonometry, central venous pressure (CVP) by a manometer and cardiac index (Cl), stroke index (Sl) and heart rare (HR) by transthoracic electrical bioimpedance. Systemic vascular resistance index (SVRI) was derived. Fluids alone failed to maintain SAP in five of 10 patients. Although Cl and CVP were maintained SVRI decreased (25 (SD 15)%; P = 0.02). Metaraminol maintained SAP in all 12 patients. The main cardiovascular change was decreased HR (15 (8)%; P = 0.0001). Ephedrine failed to maintain SAP in two of 12 patients and was accompanied by several cardiovascular changes: HR (21 (12)%; P = 0.001) increased and Sl(16 (10)%; P = 0.0001), CVP (3.8 (1.5) cm H2O; P = 0.0001) and SVRI (24 (6)%; P = 0.0001) decreased. Treatment failures resulted from failures to maintain SVRI in the fluid group and CVP and SVRI in the ephedrine group.