Early childhood is a critical period during which children develop social-emotional competence that will affect future success. Developing social-emotional assets is of importance for kindergarten children because of their concurrent cognitive and social changes as well as the experience of transitioning from home to school environment. A growing number of schools have adopted social-emotional learning (SEL) programming to focus on fostering children’s prosocial behaviors through direct instruction and engaging activities in classroom settings. However, some researchers have proposed that learning should capitalize on children’s natural interests rather than adult-determined agendas. Based on theoretical assumptions regarding potential effectiveness of child-centered group play therapy (CCGPT) as a treatment modality, I sought to explore the effects of CCGPT on social-emotional assets of kindergarten children utilizing parent and teacher reports across pretest, posttest, and one-month follow-up. Additionally, given that group sizes have been inconsistent and rarely explored across previous studies, I investigated the therapeutic aspect of group sizes in CCGPT outcome by comparing 2-member and 3-member CCGPT groups. Forty-three participants with mean age of 5.14 were recruited from three elementary schools, including 19 Hispanic, 14 Caucasian, and 10 African American. Twenty-one participants were randomly assigned to the intervention group receiving a mean of 15.32 CCGPT sessions over 8 weeks, and 22 participants were assigned to the waitlist control group. Six mixed between-within ANOVAs were conducted applying an alpha level of .05 to interpret statistical significance and η2 calculation to assess practical significance. Results indicated a statistically significant interaction effect on SEARS-P Total score, F (2, 72) = 4.533, p = .014, with medium effect size of η2 = .101. Post Hoc analyses indicated a non-statistically significant interaction effect on SEARS-P Self-Regulation/Responsibility subscale with a small effect, F (1.868, 67.248) = 1.776, p = .179, η2 = .043; a statistically significant interaction effect on SEARS-P Social Competence subscale with a medium effect, F (1.696, 61.049) = 3.413, p < .05, η2 = .079; and a statistically significant interaction effect on SEARS-P Empathy subscale with a medium effect, F (1.439, 51.79) = 4.592, p < .05, η2 = .106. Thus, participants in the CCGPT group showed a non-significant increase in the ability to self-regulate emotions and to take responsibility for actions and a significant increase in competence to interact socially with others and in empathy for others. Teacher reports did not show statistically significant results with a small effect, F (2, 70) = .917, p = .404, η2 = .013. Results on group sizes indicated that no statistical differences were detected between two-member and three-member groups with no effect, F (1.493, 25.377) = .039, p = .942, η2 = .001. Moreover, the results on follow-up supported that a certain level of the effects of CCGPT lasted one month after the intervention. Overall, children seemed to benefit from CCGPT, and CCGPT may be considered a viable treatment for enhancing kindergarten children’s social-emotional development.