Forty patients undergoing spinal anaesthesia for a variety of surgical procedures were randomly allocated to receive 3 ml of ropivacaine 5 mg ml−1 in glucose 10 mg ml−1 or 50 mg ml−1. Onset of sensory block to T10 was significantly faster (P=0.03) with the glucose 50 mg ml−1 solution (median 5 min, range 2–20 min) than with the 10 mg ml−1 solution (median 10 min, range 2–25 min). Maximum extent of cephalad spread was virtually the same in both groups (10 mg ml−1 median T6/7, range T3–T10; 50 mg ml−1 median T6, range T3–T10) with similar times to regression beyond S2 (10 mg ml−1 median 210 min, range 150–330 min; 50 mg ml−1 median 210 min, range 150–330 min). Complete motor block was produced in the majority of patients (10 mg ml−1 90%; 50 mg ml−1 85%) and the time to complete regression was the same in both groups (median 120 min, range 90–210 min). A block adequate for the projected surgery was achieved in all patients.