We use panel data to estimate multinomial logistic regressions for the effect of household composition, car ownership, distance between family members, and degree of urbanization on mode choice for family visits in the Netherlands. Based on Mundlak’s formulation our model accounts for unobserved heterogeneity and differentiates within-individual from between-individual variation. With respect to household composition, we find that living with a partner and having a child under six years old is negatively associated with the likelihood of using public transport for family visits. Number of children is not associated linearly with mode choice. Walking and cycling is mainly associated with distance between family members: the shorter the distance the higher the likelihood of using slow-modes instead of a car. Those travelling between areas of high degree of urbanization have a higher likelihood of using public transport relative to using cars. Car ownership is negatively associated with all other modes. Using a car for commuting is also found to be negatively associated with other modes for family visits.