Background: Breastfeeding is a modifiable pathway contributing to infant health. Objectives: To estimate trajectories of breastfeeding and the association between trajectories of breastfeeding and infant health. Methods: We conducted a prospective longitudinal cohort study (2014-2020, Canada) with 3217 cohabitating English-literate mother-father-infant triads with internet or telephone access. Women with major psychiatric conditions or suicidality were excluded. Data were collected online at baseline (90th percentile). We estimated trajectories of breastfeeding using latent class growth analysis. Associations were estimated using adjusted linear mixed models. Results: Most included mothers (n=2548) were of Canadian birth (2078/2548, 82%) and graduated from university (1920/2548, 75%); few reported unemployment (222/2548, 9%) or prenatal depression (240/2548, 9%). We identified two trajectories of breastfeeding: Constant Breastfeeding (1522/2548, 60%) and Decreasing Breastfeeding (1026/2548, 40%). No statistically significant differences were observed between classes for either the risk of challenging behaviors (RR -0.006 95% CI -0.028 to 0.016) or the risk of emergent visits (RR -0.015 95% CI -0.053 to 0.023). Conclusions: In this study, discontinued breastfeeding was not associated with infant health. Future studies should focus on 6-month exclusive breastfeeding, a modifiable therapeutic infant feeding target.