Background/Context: State-funded prekindergarten has expanded rapidly in the U.S., and there are expectations that it will have longer-term positive effects on later academic and nonacademic outcomes (e.g., Phillips et al., 2017). There is strong evidence of pre-k effects on measures of kindergarten readiness (e.g., Gormley et al., 2005; Weiland & Yoshikawa, 2013); however, there is not strong evidence about longer-term effects. This follow-up study through 6 grade of a subsample of students from a large randomized control trial of the Tennessee Voluntary Pre-K Program (TN-VPK) continued to investigate the long-term effects of the program on children's nonacademic outcomes. To date, the findings of the study have shown significant, positive immediate effects of TN-VPK on academic outcomes, but null or negative effects on long-term academic and non-academic outcomes through 6 grade (Authors, 2018, under revised review). This is the first well-controlled experimental study of a state pre-k program with follow-up into middle school. Purpose/Objective/Research Question: We investigated whether participation in TN-VPK had significant effects on conscientiousness, school engagement, delinquency, problem behaviors, peer relationships, and executive function in 4 through 6 grade. Setting: TN-VPK is a statewide voluntary pre-k program that prioritizes children eligible for the federal free or reduced price lunch programs and serves over 18,000 children with programs in all but a few of the school districts in the state. Population/Participants/Subjects: The current study follows a subsample of 763 children randomly assigned to participate or not in TN-VPK in 2010 who were part of a larger randomized trial. All children were from low-income households; about 47% were male, 48% were white, 27% were Black, 24% were Hispanic, and 26% were non-native English speakers. Intervention/Program/Practice: TN-VPK is a typical pre-k program and requires a licensed teacher and an aide in each classroom, a maximum class size of 20, a state-approved curriculum, and a minimum instructional time of 5.5 hours per day during the school year. When the program began, it met 9 of the 10 standards advocated until recently revised by the National Institute of Early Education Research (Barnett et al., 2009). Research Design: The larger randomized trial involved 79 over-subscribed schools in which cohorts of pre-k applicants were randomized to offers of admission or waitlist status during school years 2009-10 and 2010-11. This resulted in 111 separate school-level randomizations and a total sample of 2990 children. In 2015, we designed a follow-up intensive subsample of children (FISS) to support interviews with students, parents, and teachers. The selection of this sample focused on school lists from the original randomization with high proportions of children with earlier parental consent and a good balance between the numbers of children in the treatment and control conditions. Based on inclusion criteria and parental consent, the resulting subsample consisted of 763 students with 378 in the intent-to-treat (ITT) group offered admission and 385 in the control group of waitlisted children not offered admission. The treatment-on-treated (TOT) sample consisted of 449 VPK participants and 314 nonparticipants. Data Collection and Analysis: Students and their parents and teachers were interviewed in 4 through 6 grade. We created latent variable factor scores that combined data across measures and respondents. Missing values were estimated as the mean of 25 multiple imputations. Principal axes factor analyses were applied to the multiple measures from the respective multiple respondents for each latent construct (Costello & Osborne, 2005). We created factor scores for the resulting dominant factors using Bartlett's technique (DiStefano et al., 2009). These factor scores served as the latent variable measures for the constructs of school engagement, executive function, conscientiousness, peer relationships, problem behaviors, and delinquency. The measures and respondents represented in each factor are listed in Table 1. Hierarchical linear models with children nested within randomized lists nested within districts were used to test the effects of TN-VPK, and demographic covariates were included in all models. ITT condition was used as a predictor in the primary analyses, and the results of each ITT analysis were then used to estimate TOT effects using two-stage least-squares instrumental variable regression models. To allow estimation of TN-VPK effects for students with the statewide demographic pro le, weighting functions were created to adjust for differences between the FISS sample and the statewide pro le. Findings/Results: There were no statistically significant differences between conditions for school engagement and delinquency, and there were no statistically significant differences in 4 grade (Tables 2-4). However, in 5 grade there was a statistically significant difference on the unweighted values for conscientiousness favoring the control group, and this difference was marginally significant in 6 grade (Tables 3 and 4). The executive function scores also favored the control group, a difference that was statistically significant across the weighted and unweighted analyses in 5 and 6 grade (Tables 3 and 4). Also, children who did not attend TN-VPK showed more positive peer relationships in 6 grade than children who attended, a statistically significant difference in the weighted analysis and marginally significant in the unweighted one (Table 4). There was also a marginally significant difference favoring the treatment group on problem behaviors in 5 grade (Table 3). Conclusions: This randomized study of a state pre-k program provides a cautionary tale about what should be expected from such programs, especially regarding long-term effects. One possible explanation for these mostly negative findings is that center-based care could be preventing children from developing the internal self-control necessary for long-term development. Classrooms of 20 4-year-olds require behavioral control exerted by adults. Studies demonstrate that teachers in these circumstances often display a at to negative affect (Coelho et al., 2021; Farran et al., 2017), one that could lead to children developing negative attentional biases. Negative attentional biases have been associated with increased reactivity to later stressors (Todd et al., 2012). The long-term negative outcomes on behavior for children in group care have been found in both small experimental studies and broad-based population studies. Determining their etiology and creating classroom practices that yield different outcomes is critical.