Most evidence-based home visiting models are designed to support families from pregnancy through a child’s second birthday, though programs often struggle to retain families for this long. Previous research on client and program factors that predict attrition has produced mixed results, which may be partly because attrition is typically conceptualized as a homogeneous phenomenon. The current study sampled 991 women who received home visiting services from one of 26 agencies in a statewide network of evidence-based programs. Participants who remained in services were compared to three types of early leavers: those who communicated their intent to leave (active attrition), those whose cases closed due to non-participation (passive attrition), and those who moved from the service area. Within a year of enrollment, 42% of women exited services. Cox regression results suggested no differences in the timing of service exit among the three attrition types. Multinomial analyses revealed that, when compared to participants who remained in services, active leavers were more likely to be married or cohabitating, while passive leavers were more likely to be younger, African American, unemployed, and to have a home visitor with low job satisfaction. Participants who moved were less likely to be Latina and employed. An early pattern of inconsistent attendance was the strongest predictor of active and passive withdrawal. Rates of attrition varied by home visiting model, though inconsistent attendance was a robust predictor of passive attrition across models. This study underscores the need to scrutinize service duration as a metric of success in home visiting.