We randomized women having elective Caesarean section to receive either no preload (control group, n=33) or 4% gelatin solution (Gelofusine) 15 ml kg–1 (colloid group, n=35) i.v. before spinal anaesthesia. Intravenous metaraminol was titrated at 0.25–0.75 mg min–1 to maintain systolic arterial pressure (SAP) in the target range 90–100% of baseline after the spinal injection. The control group required more vasopressor in the first 10 min [median 1.7 (range 0–2.9) mg vs 1.4 (0–2.8), P=0.02] at a greater maximum infusion rate [0.5 (0–0.75) vs 0.25 (0–0.5) mg min–1, P=0.0005] and had a lower minimum SAP [90 (51–109) vs 101 (75–127) mm Hg, P=0.006] than the colloid group. Nausea was less frequent in the colloid group (6 vs 24%) but neonatal outcome was similar in the two groups. Colloid preload improved haemodynamic stability but did not affect neonatal outcome when arterial pressure was maintained with an infusion of metaraminol during spinal anaesthesia for Caesarean section.