In homozygous sickle cell (SS) disease, skin cooling is a common precipitating factor of the painful crisis which is associated with avascular necrosis of active bone marrow. Since skin cooling does not directly induce sickling, we have investigated the nature of the reflex vascular responses to mild cooling in SS patients in a steady state of the disease and compared them with their history of painful crises. 2. Experiments were performed in Jamaica on 60 male SS patients and 30 matched control subjects with normal haemoglobin (AA) genotype. Forearm blood flow (FBF) was measured by venous occlusion plethysmography and mean arterial pressure (MAP) by a Finapres device: forearm vascular resistance (FVR) was calculated as MAP/FBF. Cutaneous erythrocyte flux in forearm and hand was monitored by a laser Doppler meter. The contralateral hand was immersed in cool water at 16 degrees C for 2 min, 6 times, at random intervals of 0.5-3 min. 3. The first cool immersion evoked an increase in MAP, cutaneous vasoconstriction and a net increase in FVR in both AA and SS subjects. However, the direction of change in FVR varied between individuals such that 18 AA subjects showed an increase in FVR (constrictor group) while 12 showed a decrease in FVR, indicating vasodilatation in forearm muscle (dilator group). In contrast, 50 SS subjects showed an increase in FVR and only 10 showed a decrease in FVR. The proportion of subjects who showed net vasoconstriction was significantly greater in the SS than in the AA group (83% versus 60%, P = 0.03, chi 2 test). 4. By the sixth cool stimulus, the ‘dilator’ group of AA subjects showed no change in FVR while the ‘dilator’ group of SS patients showed an increase in FVR. We suggest that forearm muscle vasodilatation was the characteristic component of the alerting/defence response to novel or noxious stimuli which habituates on repetition. 5. In the whole group of SS patients, baseline values of cutaneous vascular resistance and FVR increased between stimuli, indicating persistent vasoconstriction, and the sixth cool stimulus still evoked cutaneous vasoconstriction and a net increase in FVR. In contrast, AA subjects showed an increase in baseline FVR between stimuli, but the sixth cool stimulus had no significant effect on cutaneous vascular resistances, or FVR. 6. In SS patients there were no associations between the direction of change in FVR evoked by the first cool stimulus and forearm circumference or skinfold thickness, concentrations of haemoglobin or fetal haemoglobin. However, the frequency of painful crises was significantly greater in the ‘constrictor’ group than in the ‘dilator’ group (0.36 versus 0.12/year, P = 0.04, Mann-Whitney test). 7. These results indicate that the primary reflex vasoconstrictor response evoked by mild cooling is stronger and more persistent in SS patients than in AA subjects and is particularly strong in SS patients who are most prone to painful crises. The results are consistent with the hypothesis that skin cooling may precipitate the painful crisis by causing reflex vasoconstriction in muscle, and possibly in bone marrow, so diverting blood flow away from the active marrow.