Data from the Demographic and Health Surveys (DHS) for 5 Latin American countries (7 data sets) were used to explore the feasibility of creating a composite feeding index and to examine the association between feeding practices and child height-for-age Z-scores (HAZ). The variables used for the index were as follows: current breast-feeding, use of complementary foods and liquids in the past 24 h, frequency of use over the past week and feeding frequency. The index was made age specific for 6- to 9-, 9- to 12- and 12- to 36-mo-old age groups, and age-specific feeding terciles were created. Bivariate analyses showed that feeding practices were strongly and significantly associated with child HAZ in all 7 data sets, especially after 12 mo of age. Differences in HAZ between child feeding terciles remained significant after controlling for potentially confounding influences, for all countries except Bolivia. Multiple regression analyses also revealed that better feeding practices were more important for children of lower, compared with higher socioeconomic status (in Colombia 1995 and Nicaragua 1998); among children of Ladino (Spanish speaking) compared with indigenous origin (in Guatemala 1995); and among children whose mothers had primary schooling compared with mothers with no schooling, or mothers with higher than primary school level (Peru 1996). The data available in DHS data sets can thus be used effectively to create a composite child feeding index and to identify vulnerable groups that could be targeted by nutrition education and behavior change interventions.