Special education placements for more than 69,000 preschoolers with disabilities were examined within and across five southern states. Data were gathered from the 2007 December 1st Child Count reported to the U.S. Department of Education. All states examined offered state-funded prekindergarten programs. Analyses compared disproportionate representation in placements (three levels of inclusion) across child ethnicity and state of residence. Disproportionate representation indexes were calculated using risk index (RI) and risk ratio (RR). Results reveal that a full inclusion setting was used most frequently, followed by no known inclusion, and finally partial inclusion. Overall, although individual state variation was great, no disproportionate representation occurred in full or partial inclusion settings, whereas children classified as American Indian were underrepresented in the no known inclusion setting (RR = 0.22). Inspection of individual RI by state found significant differences (Wilks's [lambda] = 0.001, F = 10.354, p = 0.001, [eta][superscript 2] = 0.902) between states for full (p less than 0.05, [eta][superscript 2] = 0.902), partial (p less than 0.001, [eta][superscript 2] = 0.873), and no known inclusion (p less than 0.05, [eta][superscript 2] = 0.699) settings. Policy implications of the data are discussed. (Contains 5 tables.)