This study investigates heterogeneity in the trajectories of global self-esteem (GSE) in adolescence and the relations between these trajectories and students' school experiences and climate perceptions encompassing interpersonal, organizational and instructional components of school life. Methodologically, this study illustrates the usefulness of growth mixture modeling to get a fuller representation of GSE development, and how to obtain proper student-level effects when there are multiple schools, but not enough to support multilevel analyses. Data from the Montreal Adolescent Depression Development Project, a four-year follow-up of over 1000 adolescents who completed the Rosenberg Self Esteem Inventory each year, was used. Growth mixture analyses converged on a 4-class model: (a) A high and increasing trajectory (16.4%), (b) a stable low trajectory (13.7%), (c) a stable moderate trajectory (52.6%), (d) an initially moderate trajectory that shows a marked increase and reach the level of the high trajectory (17.3%). This last trajectory clearly illustrate that self esteem is not a necessarily stable phenomenon and can improve over the course of adolescence. To validate the extracted trajectories, predictors related to students' school experiences were added to the model: grades, relationships with teachers (warmth and support), school victimization, loneliness, and perceptions of the school climate (relational, security, bonding, educative, and justice). Few of these variables predicted membership in specific classes, suggesting that global self-esteem is mostly determined by factors unrelated to school experiences. The variables that did predict membership (e.g., loneliness, grades, relationships with teacher) suggest that school experiences affect individual students' through micro, rather than macro (i.e. climate) processes. However, if students' perceptions of their school climate showed few direct relations with membership in specific trajectories, many facets of school climate perceptions interacted with gender in predicting trajectory-membership, showing that boys and girls needs at school are different. The discussion will also underscore some methodological implications and limitations of the present results, and directions for future studies.