Background: Early school leaving (ESL) is a significant and complex problem for most educational systems. Research has analysed this problem from a number of different perspectives but has been mainly focused on a specific set of variables that may influence ESL. Purpose: This study sought to identify the variables that influence ESL in compulsory education, provide a global perspective and develop a comprehensive model to enable the visualisation of different ways in which a student can be at risk of ESL. Design and method: A content analysis of 32 reviews published between 2006 and 2016 was carried out, and data were coded qualitatively. Findings: In total, 122 variables connected with ESL were grouped into two differentiated clusters: a non-academic cluster with three factors (Individual, Family and Friendship) and an academic cluster with four factors (Student, School, Teacher and Classmates). Based on the variables analysed, a model was created that presents, in a synthetic and visual form, all variables related to ESL, grouped in two clusters and seven factors. The largest groups of variables related to ESL are those linked to the person, as an individual and a student, and those linked to the family. This classification provides an idea of the complexity of ESL as well as the high number of variables that should be considered in designing efficient strategies to provide solutions to it. Conclusions: This study points to the need to approach the problem by taking into account all possible perspectives, as suggested by the proposed model, thereby taking into consideration all the variables that can contribute to ESL.