Using data from the Fragile Families and Child Wellbeing Study, we examined patterns of nonresident father involvement 1 and 3 years after a nonmarital birth (N = 893). Cluster analyses were used to determine patterns of involvement across different father behaviors. About half of fathers displayed low involvement when children were 1 and 3 years old, one fourth of fathers maintained high involvement, and equal remaining proportions increased or decreased involvement over time. Multinomial logistic analyses indicated that better relationships between parents were associated with consistently high versus low involvement. Better relationships with each others' extended family also predicted remaining highly involved and increasing involvement over time. Parents' romantic relationship status was closely associated with patterns of involvement.