At present, researchers know little about the long-term effects of prenatal substance exposure on children's behaviors and functioning. Some research has demonstrated a mild, subtle effect of drug exposure over the near term for children remaining in biological families. However, the public perception of dramatic consequences for prenatal drug exposure may have serious implications for the willingness of parents to adopt drug-exposed children. Also, little is known about the long-term correlates of adoption from foster care versus independent or other forms of adoption. The current study draws data from the 4th wave of the California Long-Range Adoption Study (CLAS), the only longitudinal adoption study with a significant number of former foster children. Data from the 3rd wave of CLAS (8 years) showed that adopted drug-exposed children experienced modestly higher levels of problems than adopted children not exposed to drugs. This difference is also true of foster children, a situation that requires a disentangling of these effects. In the present study, the authors constructed a two-level growth curve model to analyze the effects of substance exposure on children's behavior (N=332) at 2, 4, 8, and 13 years postadoption, using a reduced form of the Behavior Problems Index (BPI) as an outcome variable. This model is a variant of hierarchical linear modeling (HLM), in which waves (1st level) are nested within children (2nd level). Other 2nd level covariates include prenatal substance exposure, adoption from foster care, age at adoption, and child's gender; all two-way interactions were tested. Results indicated that behavior problems in general increased over time (p <.01), and that substance-exposed children consistently experienced higher levels of behavior problems than their non-exposed counterparts (p <.05). Age at adoption had no overall effect on increased problems. Children adopted from foster care had higher levels of problems than those not adopted from foster care (p <.05). Boys were more likely to experience higher levels of problems than girls (p <.01). Interaction effects were present for baseline age and wave (p<.01) and for baseline age and adoption from foster care (p<.01). The presence of these effects indicates a differential change rate in BPI scores based on children's age of adoption. While older children are no more likely than younger children to have higher levels of problems, younger children experience a higher rate of worsening change over time when compared to older children. Foster care children are faring worse than non foster children overall; however, for children adopted older than 6, the change rates between foster and non foster children are similar. Drug-exposed children have higher levels of behavior problems with a consistent rate of change over time in BPI scores. Because non foster children adopted at younger ages have a smaller rate of change, and this rate increases as age at adoption increases, these findings buttress the importance of early adoption from foster care. The findings will also be used to discuss the use of growth curve modeling and interpretations of interactions with wave—topics seldom discussed in social work research.