Background: Anthracycline-taxane chemotherapy for early-stage breast cancer substantially improves survival compared with no chemotherapy. However, concerns about short-term and long-term side-effects of anthracyclines have led to increased use of taxane chemotherapy without anthracycline, which could compromise efficacy. We aimed to better characterise the benefits and risks of including anthracycline, and the comparative benefits of different anthracycline-taxane regimens. Methods: We did an individual patient-level meta-analysis of randomised trials comparing taxane regimens with versus without anthracycline, and updated our previous meta-analysis of anthracycline regimens with versus without taxane, as well as analysing 44 trials in six related comparisons. We searched databases, including MEDLINE, Embase, the Cochrane Library, and meeting abstracts to identify trials assessing anthracycline and taxane chemotherapy. Adjuvant or neoadjuvant trials were eligible if they began before Jan 1, 2012. Primary outcomes were breast cancer recurrence and cause-specific mortality. Log-rank analyses yielded first-event rate ratios (RRs) and CIs. Findings: 28 trials of taxane regimens with or without anthracycline were identified, of which 23 were deemed eligible, and 15 provided data on 18 103 women. Across all 15 trials that provided individual data, recurrence rates were 14% lower on average (RR 0·86, 95% CI 0·79-0·93; p=0·0004) with taxane regimens including anthracycline than those without. Non-breast cancer deaths were not increased but there was one additional acute myeloid leukaemia case per 700 women treated. The clearest reductions in recurrence were found when anthracycline was added concurrently to docetaxel plus cyclophosphamide versus the same dose of docetaxel plus cyclophosphamide (10-year recurrence risk 12·3% vs 21·0%; risk difference 8·7%, 95% CI 4·5-12·9; RR 0·58, 0·47-0·73; p<0·0001). 10-year breast cancer mortality in this group was reduced by 4·2% (0·4-8·1; p=0·0034). No significant reduction in recurrence risk was found for sequential schedules of taxane plus anthracycline when compared with docetaxel plus cyclophosphamide (RR 0·94, 0·83-1·06; p=0·30). For the analysis of anthracycline regimens with versus without taxane, 35 trials (n=52 976) provided individual patient data. Larger recurrence reductions were seen from adding taxane to anthracycline regimens when the cumulative dose of anthracycline was the same in each group (RR 0·87, 0·82-0·93; p<0·0001; n=11 167) than in trials with two-fold higher cumulative doses of non-taxane (mostly anthracycline) in the control group than in the taxane group (RR 0·96, 0·90-1·03; p=0·27; n=14 620). Direct comparisons between anthracycline and taxane regimens showed that a higher cumulative dose and more dose-intense schedules were more efficacious. The proportional reductions in recurrence for taxane plus anthracycline were similar in oestrogen receptor-positive and oestrogen receptor-negative disease, and did not differ by age, nodal status, or tumour size or grade. Interpretation: Anthracycline plus taxane regimens are most efficacious at reducing breast cancer recurrence and death. Regimens with higher cumulative doses of anthracycline plus taxane provide the greatest benefits, challenging the current trend in clinical practice and guidelines towards non-anthracycline chemotherapy, particularly shorter regimens, such as four cycles of docetaxel-cyclophosphamide. By bringing together data from almost all relevant trials, this meta-analysis provides a reliable evidence base to inform individual treatment decisions, clinical guidelines, and the design of future clinical trials.