The WHO/UNICEF Baby‐friendly Hospital Initiative has been shown to increase breastfeeding rates, but uncertainty remains about effective methods to improve breastfeeding in community health services. The aim of this pragmatic cluster quasi‐randomised controlled trial was to assess the effectiveness of implementing the Baby‐friendly Initiative (BFI) in community health services. The primary outcome was exclusive breastfeeding until 6 months in healthy babies. Secondary outcomes were other breastfeeding indicators, mothers' satisfaction with the breastfeeding experience, and perceived pressure to breastfeed. A total of 54 Norwegian municipalities were allocated by alternation to the BFI in community health service intervention or routine care. All mothers with infants of five completed months were invited to participate (n = 3948), and 1051 mothers in the intervention arm and 981 in the comparison arm returned the questionnaire. Analyses were by intention to treat. Women in the intervention group were more likely to breastfeed exclusively compared with those who received routine care: 17.9% vs. 14.1% until 6 months [cluster adjusted odds ratio (OR) = 1.33; 95% confidence interval (CI): 1.03, 1.72; P = 0.03], 41.4% vs. 35.8% until 5 months [cluster adjusted OR = 1.39; 95% CI: 1.09, 1.77; P = 0.01], and 72.1% vs. 68.2% for any breastfeeding until 6 months [cluster adjusted OR = 1.24; 95% CI: 0.99, 1.54; P = 0.06]. The intervention had no effect on breastfeeding until 12 months. Maternal breastfeeding experience in the two groups did not differ, neither did perceived breastfeeding pressure from staff in the community health services. In conclusion, the BFI in community health services increased rates of exclusive breastfeeding until 6 months. © 2015 Blackwell Publishing Ltd