We conduct a field experiment to evaluate the extent to which dietary habits are malleable early on in childhood and later in life. We implement two treatments one that targets what people eat, the other that targets the timing and frequency of food intake. 285 low income families with young children were recruited and assigned either to a control group or one of the two treatments, each of them lasting for 12 consecutive weeks. In one treatment, families received food groceries at home for free for 12 weeks and were asked to prepare five specific healthy meals per week. In the other treatment, families were simply asked to reduce snacking and eat at regular times. We collected a range of measures of food preferences, dietary intake, as well as BMI and biomarkers based on blood samples. We find evidence that children's BMI distribution shifted significantly relative to the control group, i.e. they became relatively "thinner". We also find some evidence that their preferences have been affected by both treatments. On the other hand, we find little evidence of effects on parents. We conclude that exposure to a healthy diet and regularity of food intake possibly play a role in shaping dietary habits, but influencing dietary choices later on in life remains a major challenge.