Background: Childhood internalizing and externalizing problems are closely related and often co-occur. Directional models have been employed to test how these problems are related, while few studies have tested a third variables model. Objective: This study investigates whether internalizing and externalizing problems are reciprocally or unidirectionally related, whether these relations can be explained by third variables, and how these relations are associated with onset and stability. Methods: A community sample of 1,434 children aged 5.08 (SD = 1.25) and their mothers participated in two 1-year interval data waves. Internalizing and externalizing problems were examined with the Strengths and Difficulties Questionnaire. Results: Using latent cross-lagged modeling, externalizing problems were found not be related to subsequent internalizing problems, or vice versa. These results were also found when controlling for inadequate parenting, parenting stress, maternal health and social preference. When taking problem level into account, externalizing problems were related to stability of clinical level internalizing problems, even when controlling for third variables inadequate parenting, parenting stress, maternal mental health and social preference. Conclusions: Strong autoregressive paths for internalizing and externalizing problems were found. Internalizing and externalizing problems do not seem to influence each other over time in the community sample. When investigating relations among internalizing and externalizing problems, it seems to be important to take problem level into account.