This pilot study examined the utility of a practice-based coaching (PBC) intervention in Swedish elementary schools designed to support the classroom climate and benefit students. The intervention involved activities, self-assessment, observation, and coaching to the teachers, with the aim of improving the classroom climate and students' self-concept, prosocial behavior, well-being, and academic achievement. The design was a cluster randomized trial with the school as the assignment unit. Participants in the intervention were 66 students in four classes, and in the control, four classes with 77 students. The duration of the intervention was five months. An autoregressive structural equation model was estimated. The correlations between the latent variables at pre-test ranged from low to high. There were significant standardized path coefficients concerning all the latent variables pre- and post-test. The estimates of the intervention at the post-test were non-significant. The results suggested a longer duration of the intervention with more frequent coaching sessions, which may enhance its effect.