Background: The optimal way to schedule AIs and/or Tam in the adjuvant treatment of early breast cancer remains uncertain. Methods: ITT meta-analysis of individual patient data on 36 889 post-menopausal women with ER-positive invasive breast cancers in randomised trials of [A] Continuous AI (5yrs) vs Tam (5yrs); [B] Sequential Tam then AI (2-3yrs of Tam then 2-3 yrs AI) vs Tam (5yrs); [C] Continuous AI (5yrs) vs Sequential Tam then AI (5yrs). Results: [A] Fewer women had breast cancer recurrence with Continuous AI than Tam (827/4,970 vs 964/4,915, p<0.0001) and fewer died of breast cancer: 504 vs 562; rate ratio (RR) 0.86 [0.76-0.97], p=0.014. Recurrence RRs were: 0.66 during yrs 0-1 [95%CI 0.54-0.80], 0.75 during yrs 2-4 [0.64-0.88] and 0.90 in yrs 5+ [0.79-1.04]. [B] Recurrence was also lower with Sequential Tam then AI than with Tam alone (753/5,909 vs 863/5,889, p=0.0001) as was breast cancer mortality (361 vs 428 deaths; RR 0.84 [0.73-0.97], p=0.015). Recurrence RRs were 0.56 during yrs 2-4 [0.46-0.67] and 0.97 in yrs 5+ [0.86-1.09]. [C] In trials comparing Continuous AI versus Sequential Tam then AI, recurrence was lower with AI than Tam during yrs 0-1; RR 0.75 [0.62-0.89], but similar during yrs 2-4 (0.99 [0.85-1.15]), when both groups received AI, and in yrs 5+ (0.96 [0.76-1.21]) after treatment completion; overall, there were fewer recurrences with Continuous AI than Sequential Tam then AI (705/6,422 vs 764/6,377, RR 0.90 [0.81-1.00]; 5yr recurrence 9.6% vs 10.7%, p=0.042) and fewer breast cancer deaths (395 vs 432; 0.89 [0.77-1.02], p=0.097). In the 3 comparisons, proportional recurrence reductions did not differ much by age, nodal status, tumour grade, or PR status and, overall, fewer endometrial cancers (0.2% vs 0.6%, RR=0.37 [0.27-0.51]) but more fractures (8.1% vs 5.9%, RR=1.40 [1.27-1.53]) were seen with AIs than Tam; non-breast deaths were similar. Conclusions: AIs, in either Continuous or Sequential regimens, are even more effective than Tam monotherapy in preventing recurrence and breast cancer death, despite substantial cross-over from Tam to AI in some studies. Recurrence reductions are seen mainly while treatments differ, hence somewhat fewer recurrences with Continuous AI than Sequential Tam then AI.