BACKGROUND: Analyses of some randomised trials show that calcium-channel blockers reduce the risk of stroke more than expected on the basis of mean blood pressure alone and that beta blockers are less effective than expected. We aimed to investigate whether the effects of these drugs on variability in blood pressure might explain these disparities in effect on stroke risk. METHODS: The Anglo-Scandinavian Cardiac Outcomes Trial Blood Pressure Lowering Arm (ASCOT-BPLA) compared amlodipine-based regimens with atenolol-based regimens in 19 257 patients with hypertension and other vascular risk factors and the Medical Research Council (MRC) trial compared atenolol-based and diuretic-based regimens versus placebo in 4396 hypertensive patients aged 65-74 years. We expressed visit-to-visit variability of blood pressure during follow-up in the two trials as standard deviation (SD) and as transformations uncorrelated with mean blood pressure. For ASCOT-BPLA, we also studied within-visit variability and variability on 24 h ambulatory blood-pressure monitoring (ABPM). RESULTS: In ASCOT-BPLA, group systolic blood pressure (SBP) SD was lower in the amlodipine group than in the atenolol group at all follow-up visits (p<0.0001), mainly because of lower within-individual visit-to-visit variability. Within-visit and ABPM variability in SBP were also lower in the amlodipine group than in the atenolol group (all p<0.0001). Analysis of changes from baseline showed that variability decreased over time in the amlodipine group and increased in the atenolol group. The lower risk of stroke in the amlodipine group (hazard ratio 0.78, 95% CI 0.67-0.90) was partly attenuated by adjusting for mean SBP during follow-up (0.84, 0.72-0.98), but was abolished by also adjusting for within-individual SD of clinic SBP (0.99, 0.85-1.16). Findings were similar for coronary events. In the ABPM substudy, reduced variability in daytime SBP in the amlodipine group (p<0.0001) partly accounted for the reduced risk of vascular events, but reduced visit-to-visit variability in clinic SBP had a greater effect. In the MRC trial, group SD SBP and all measures of within-individual visit-to-visit variability in SBP were increased in the atenolol group compared with both the placebo group and the diuretic group during initial follow-up (all p<0.0001). Subsequent temporal trends in variability in blood pressure during follow-up in the atenolol group correlated with trends in stroke risk. INTERPRETATION: The opposite effects of calcium-channel blockers and beta blockers on variability of blood pressure account for the disparity in observed effects on risk of stroke and expected effects based on mean blood pressure. To prevent stroke most effectively, blood-pressure-lowering drugs should reduce mean blood pressure without increasing variability; ideally they should reduce both.