Background: Child malnutrition, a leading cause of death and disability worldwide, is particularly severe in Madagascar, where 47% of children under 5 years are stunted (low height-forage) and 8% are wasted (low weight-for-height). Widespread poverty and a weak health system have hindered attempts to implement life-saving malnutrition interventions in Madagascar during critical periods for growth faltering. Objective: This study aimed to shed light on the most important factors associated with child malnutrition, both acute and chronic, and the timing of growth faltering, in Ifanadiana, a rural district of Madagascar. Methods: We analyzed data from a 2014 district-representative cluster household survey, which had information on 1175 children ages 6 months to 5 years. We studied the effect of child health, birth history, maternal and paternal health and education, and household wealth and sanitation on child nutritional status. Variables associated with stunting and wasting were modeled separately in multivariate logistic regressions. Growth faltering was modeled by age range. All analyses were survey-adjusted. Results: Stunting was associated with increasing child age (OR = 1.03 (95%CI 1.02-1.04) for each additional month), very small birth size (OR = 2.32 (1.24-4.32)), low maternal weight (OR = 0.94 (0.91-0.97) for each kilogram, kg) and height (OR = 0.95 (0.92-0.99) for each centimeter), and low paternal height (OR = 0.95 (0.92-0.98)). Wasting was associated with younger child age (OR = 0.98 (0.97-0.99)), very small birth size (OR = 2.48 (1.23-4.99)), and low maternal BMI (OR = 0.84 (0.75-0.94) for each kg/m 2). Height-forage faltered rapidly before 24 months, then slowly until age 5 years, whereas weight-for-height faltered rapidly before 12 months, then recovered gradually until age 5 years but did not reach the median. Conclusion: Intergenerational transmission of growth faltering and early life exposures may be important determinants of malnutrition in Ifanadiana. Timing of growth faltering, in the first 1000 days, is similar to international populations; however, child growth does not recover to the median.