BACKGROUND: Azithromycin mass drug administration (MDA) could reduce child mortality. However, macrolide resistance, which has generally been reported to develop after whole-community MDA for trachoma control, is a concern, and it has less commonly been studied in the context of treating children to reduce mortality. Here, we report on macrolide resistance after biannual azithromycin MDA at the Malawi site of the MORDOR study. METHODS: In the MORDOR cluster-randomised trial in Malawi, 30 communities in Mangochi District were randomly selected. Communities were randomly assigned to receive azithromycin or placebo by simple randomisation without stratification. Children aged 1-59 months were administered azithromycin 20 mg/kg or placebo as an oral suspension biannually for a total of four treatments in 2015-17. 1200 children (40 children per community) were randomly selected for nasopharyngeal swabs at baseline, 12 months (6 months after the second treatment visit), and 24 months (6 months after the fourth treatment visit). Samples were processed to culture Streptococcus pneumoniae. The primary outcome was the proportion of S pneumoniae isolates exhibiting macrolide resistance at 12 months and 24 months, assessed in the intention-to-treat population. The study is registered with ClinicalTrials.gov, NCT02048007. FINDINGS: At baseline, 3467 (76%) of 4541 eligible children in the azithromycin group and 3107 (72%) of 4308 eligible children in the placebo group were treated. 564 nasopharyngeal swabs were taken from the azithromycin group and 563 from the placebo group, with similar numbers of swabs taken at 12 months and 24 months. In both groups at baseline, carriage of S pneumoniae was greater than 85% and the proportion of strains resistant to macrolides was 28%. At the 12-month follow-up, macrolide resistance was higher in the azithromycin group (36·9%, 95% CI 32·5-41·2) than in the placebo group (21·6%, 17·7-25·4; OR 2·26, 95% CI 1·46-3·49; p=0·0002). At 24 months, macrolide resistance remained higher in the azithromycin group (43·9%, 39·2-48·5) compared with placebo (32·8%, 28·5-37·1; OR 1·66, 1·15-2·40; p=0·0069). INTERPRETATION: These findings support previous evidence from trachoma MDA programmes and suggest that monitoring of macrolide resistance should remain a key component of azithromycin interventions for reducing child mortality. FUNDING: Bill & Melinda Gates Foundation.