Objective. It is widely believed that only a minority of vulnerable children and adolescents receive any mental health services. Although health care disparities associated with sociodemographic characteristics are well known, almost no information exists about another potentially important source of disparity for children: How does state of residence affect mental health service use? Methods. Observational analysis was conducted using the 1997 and 1999 waves of the National Survey of America’s Families (N 45 247 children aged 6–17), a population survey fielded in 13 states and a smaller geographically dispersed sample. We studied 4 dependent variables: 1) use of any mental health services and number of visits among users; 2) need for mental health care, based on 6 items from the Child Behavior Checklist; 3) unmet need (no services among children with identified need); and 4) need among users of mental health services. Results. Use of any mental health care differs >2-fold across states, ranging from 5% in California and Texas to >10% in Colorado and Massachusetts. The variation across states in service use and unmet need exceeds the differences across racial/ethnic groups or family income. For example, the odds ratio of unmet need in California versus Massachusetts is 3.04, compared with 2.33 between Hispanic and white children. Differences in population characteristics across states do not explain much of the observed geographic variation in mental health related outcomes for children. Perhaps the most disconcerting finding is that the differences in use are not paralleled by differences in need. Overall, there is no apparent relationship between levels of need and use of services across states. As a general rule, states with high rates of services do not have low levels of need or vice versa, although that situation exists. Alabama and Texas, for example, have higher rates of need and lower rates of use than the nation as a whole, whereas Washington state displays the opposite pattern. Even with the similar levels of need and service use, states differ in the effectiveness of their delivery system. Alabama and Mississippi have high rates of need and low levels of use, but rates of unmet need are not significantly higher in those 2 states than in the nation, whereas California, Florida, and Texas have the highest rates of unmet need. In California and Texas, children from high-income families are more likely to receive some mental health services than children from low-income families. In Alabama and Mississippi, as well as in the states with the lowest rates of unmet need (Colorado, Massachusetts, and Minnesota), the opposite is true: children from low-income families are much more likely to receive any mental health service than children from high-income families. Conclusions. Large differences from the national average across states in service use and unmet need are the rule, rather than the exception. National averages obscure large differences that can exceed the effects of race/ethnicity or income. The differences in the rates of use or unmet need are not driven by differences in the racial/ ethnic or socioeconomic makeup across states but more likely are the result of differences in state policies and health care market characteristics. These state policies and health care market characteristics can interact with sociodemographic characteristics and affect how effectively resources are used. For states such as California and Texas that have the lowest rates of mental health service use, it may be less important to raise the rates of service use than to deliver them to the children with the highest need, predominantly black and Hispanic children and children in low-income families. Pediatrics 2003;112:e308–e315. URL: http://www.pediatrics.org/cgi/ content/full/112/4/e308; unmet need, health disparities, behavioral health, state variations.