The present study examined predictors and pathways relevant to children's dosage in Early Head Start (EHS) using the dataset from the Early Head Start Family and Child Experiences Study (Baby FACES; Vogel & Boller, 2009-2012). Dosage was defined in two ways: (a) leaving EHS early, before age eligibility ended, and (b) the total length of EHS enrollment. This study first examined factors predicting children's dosage of EHS with a particular focus on parent-caregiver relationships captured when children were age 1. Second, this study examined a potential mediational role of family involvement on the association between parent-caregiver relationships and EHS dosage, controlling for child and family characteristics. Findings showed that children from families with higher risks were more likely to leave EHS early and were enrolled in EHS for less time. Children were less likely to leave EHS early when their caregivers reported positive relationships with parents when children were age 1. However, the length of EHS enrollment was not predicted by parent-caregiver relationships as reported either by parents or caregivers. Findings from mediation models showed that caregiver-reported positive parent-caregiver relationships were associated with the higher level of family involvement during EHS which, in turn, predicted both lower rates of early leaving and longer length of EHS enrollment. These results indicate that promoting initial positive parent-caregiver relationships and encouraging family involvement in EHS may be important for maximizing children's EHS dosage, which has been shown to be beneficial for the development of children in EHS. (author abstract)