The Academy of Nutrition and Dietetics released a position statement in 2012 regarding benchmarks for nutrition in child care to establish healthful eating behaviors in early childhood and prevent obesity in young children. Further, recent publications by the Institute of Medicine and national organizations have recommended specific feeding practices for child care providers to be implemented at the state-level to prevent early childhood obesity in preschool children. Although over 12 million US children in child care consuming up to 5 meals and snacks per day in such settings, little is known about child care providers' feeding practices. The purpose of this dissertation was to examine provider's feeding practices in Head Start and child care programs with the following aims 1) To assess whether providers met the Academy's benchmarks and if attainment of benchmarks varied across child care contexts (Head Start [HS], Child and Adult Care Food Program [CACFP] and non-CACFP). 2) To identify determinants of child care providers' healthful and controlling feeding practices for 2-5y-old children. 3) To identify Head Start and child care provider's motivators, barriers and facilitators for using family style meal service (FSMS) during child care mealtimes. Cross-sectional data was collected in 2011 and 2012 where 118 child care providers from 24 center-based programs (6 Head Start [HS], 11 Child and Adult Care Food Program [CACFP] funded, 7 non-CACFP) completed self-administered surveys regarding their feeding practices for 2-5-year-old children. Chi-square tests and Analysis of variance were used to determine variation in meeting benchmarks across contexts. Multi-level multivariate linear regression models were used to predict seven feeding practices- healthful (allowing children to control their food intake, role modeling healthy eating and teaching children about nutrition) and controlling (pressuring children to eat and restricting access to food for health or weight control). For aim 3, qualitative semi-structured interviews were conducted with a subset of providers based on maximum variation purposive sampling. The interviews were lead until saturation was reached and the data was coded using thematic analysis by NVivo qualitative software. (author abstract)