Breastfeeding and human milk are the gold standard for infant feeding. Studying human milk with a systems biology approach in a large longitudinal cohort is needed to understand its complexity and health implications. The Phoenix study is a multicenter cohort study focusing on the interactions of maternal characteristics, human milk composition, infant feeding practices, and health outcomes of Chinese mothers and infants. A total of 779 mother-infant dyads were recruited from November 2021 to September 2022, and 769 mother-infant dyads were enrolled in the study. Scheduled home visits took place at 1, 4, 6, and 12 months postpartum, and 696 dyads (90.5% participants) completed the 12-month visit. At each visit, maternal and infant anthropometry was assessed. Questionnaires were administered to collect longitudinal information on maternal characteristics and lifestyle, infant feeding, and health. Digital diaries were used to record maternal dietary intake, infant feeding, and stool character. Human milk, maternal feces, infant feces, and infant saliva were collected. An external pharmaceutical-level quality assurance approach was implied to ensure the trial quality. Multi-omics techniques (including glycomics, lipidomics, proteomics, and microbiomics) and machine learning algorithms were integrated into the sample and data analysis. The protocol design of the Phoenix study provides a framework for prospective cohort studies of mother-infant dyads and will provide insights into the complex dynamics of human milk and its interplay with maternal and infant health outcomes in the Chinese population.