Aim: Describe the type and amount of beverages consumed by Australian children at age 2, 3.7 and 5; investigate the longitudinal relationship between intake of sweet beverages with fruits and vegetables or milk/alternatives, and body mass index (BMI) z-score. Methods: Mothers in the NOURISH trial completed a single 24-hour recall of their child's intake, at age 2 (n = 515), 3.7 (n = 426) and 5 (n = 405). Anthropometry was measured by study staff. At each time point, proportion of children consuming at least one beverage on 24-hour recall from the following groups was determined; essential—cow's milk/alternatives, breast milk; non-essential—formula, sweet beverages. For consumers, intake (grams) of each beverage and proportion total estimated energy intake was calculated. The longitudinal relationship between intake of sweet beverages, and fruit and vegetable intake (g/kg body weight), with BMI z-score at 2, 3.7 and 5 years was examined using structural equation modelling. A second model investigated relationship with intake of milk/alternatives. Results: Sweet beverages were consumed by 38, 55 and 47% of children at each time point. Intake of sweet beverages strongly correlated between two and five years; however, intake was not associated with BMIz and did not appear to displace fruit and vegetable intake. Intake of cow's milk declined over time and was negatively associated with intake of sweet beverages at two and five years. Conclusions: Tracking of intake over time supports the notion that children's early experience of sweet tastes is a strong predictor of future intake. Limiting exposure to sweet beverages in early childhood remains an important strategy in the development of healthy food preferences and promotion of dietary quality.