This study used a person-centered approach to examine mother-daughter dyad behaviors when jointly solving addition problems during a card game. The goal was to identify maternal and child profile behaviors during the interaction as predictors of children's autonomous addition accuracy and strategy use at the end of first grade. Videotaped observations of the dyads playing addition card games (N = 91) were coded by trained observers. Latent profile analysis identified three profiles: (1) the Independent profile (52% of sample) included children who often solved independently and received little maternal support; (2) the Supported profile (35%) had children who received greater maternal support; and (3) the Dependent profile (13%) included children who received the most help from mothers. The three mother-daughter profiles differed in terms of children's: (1) independence level; (2) task mastery behaviors; and (3) strategy use. The mothers varied in terms of: (1) use of simplifying behaviors; (2) use of math fact hints; and (3) support of different types of addition strategies. OLS regression models (controlling for mother demographics and child cognitive skills) revealed that children in the Independent profile had higher addition skills at the end of first grade in comparison to children in the Dependent profile, with the Supported children in between. Children in the Independent and Supported profiles used advanced decomposition strategies more frequently in comparison to the Dependent profile children. Taken together, a person-centered approach revealed latent subgroups of maternal-child interactions that were associated with children's later addition accuracy and strategy use.