BACKGROUND Hypothermia is nearly as common, and may be as severe, during spinal and epidural anesthesia as during general anesthesia. The authors have proposed that thermoregulatory failure results when regional anesthesia increases apparent leg skin temperature to a level far exceeding actual leg skin temperature. Extensive dermatomal blocks will alter thermal input to the hypothalamus from a greater skin-surface area more than less extensive ones and thus might be expected to impair central thermoregulatory control more. Accordingly, they tested the hypothesis that reduction in the shivering threshold is directly related to the number of dermatomes blocked during spinal anesthesia. METHODS Eleven men, aged 62 +/- 6 yr (mean +/- SD), undergoing urologic surgery were studied. Ice-cold lactated Ringer's solution was administered intravenously before spinal blockade and the shivering threshold (triggering core temperature) was established. Spinal anesthesia then was induced using a randomly assigned dose of 0.5% bupivacaine (2-4 ml). Again, sufficient cold lactated Ringer's solution was given to induce shivering. Tympanic membrane, ambient and skin temperatures were measured, and extent of block was defined by loss of temperature discrimination. Presence of shivering was evaluated by a blinded observer. Mean upper-body skin and ambient temperatures, cooling rates and intravenous fluid volumes at the two thresholds were compared using paired, two-tailed t tests (P < 0.05). Linear regression defined the relationship between reduction in shivering threshold and the number of dermatomes blocked. RESULTS There were no significant differences between mean upper-body skin and ambient temperatures, cooling rates or intravenous fluid volumes at the control and spinal shivering thresholds. Spinal anesthesia reduced the shivering threshold in direct relation to the number of dermatomes blocked: delta threshold = 0.74 - 0.06 (dermatomes blocked); r2 = 0.58, P < 0.006. CONCLUSIONS Extensive spinal blockade impairs central thermoregulatory control more than less extensive blockade. Clinicians can thus anticipate more core hypothermia during extensive than during restricted spinal blockade.