Previous meta-analyses suggested that acute myeloid leukaemia induction regimens containing idarubicin (IDA) or high-dose daunorubicin (HDD) induce higher rates of complete remission (CR) than conventional-dose daunorubicin (CDD), with a possible benefit in overall survival. However, robust comparisons between these regimens are still lacking. We conducted a mixed treatment comparison meta-analysis regarding these three regimens. Mixed treatment comparison is a statistical method of data summarization that aggregates data from both direct and indirect effect estimates. Literature search strategy included MEDLINE, EMBASE, Cochrane, Scielo and LILACS, from inception until August 2013 and resulted in the inclusion of 17 trials enrolling 7258 adult patients. HDD [relative risk (RR) 1.13; 95% credible interval (CrI) 1.02–1.26] and IDA (RR 1.13; 95% CrI 1.05–1.23) showed higher CR rates than CDD. IDA also led to lower long-term overall mortality rates when compared with CDD (RR 0.93, 95% CrI 0.86–0.99), whereas HDD and CDD were no different (RR 0.94, 95% CrI 0.85–1.02). HDD and IDA comparison did not reach statistically significant differences in CR (RR 1.00; 95% CrI 0.89–1.11) and in long-term mortality (RR 1.01, 95% CrI 0.91–1.11). IDA and HDD are consistently superior to CDD in inducing CR, and IDA was associated with lower long-term mortality. On the basis of these findings, we recommend incorporation of IDA and HDD instead of the traditional CDD as standard treatments for acute myeloid leukaemia induction. The lack of HDD benefit on mortality, when compared with CDD in this study, should be cautiously addressed, because it may have been susceptible to underestimation because of statistical power limitations. Copyright © 2014 John Wiley & Sons, Ltd.