The objective of the study was to determine the causal effects of school failure (SF) among secondary school students, belonging to five public schools within the region of Girardota, Colombia, through the validation of a psychosocial model with structural equations. A total of 319 students, 25% more males, enrolled in classes between 6th and 11th year, with an average age of 14 years. Furthermore, 265 parents and 200 teachers were also included in the sample. Participants answered the questions raised in 9 instruments. Of the total number of students, 63.8% were surveyed. The instruments were subjected to a pilot test and to the judgment of experts. In order to reduce the amount of data, exploratory and confirmatory factorial analyses were used. Other techniques of multivariate analysis such as decision trees and linear regressions were also used in order to previously evaluate the relationships between the independent variables (IV) and the dependent variable (DV). Afterwards, the Full SEM was calculated, yielding a model consisting of 34 variables (10 latent and 24 observable), with the following indexes of goodness of fit: CMIN/DF = 1.146, p = .058, IFI = 0.974, TLI = .970, CFI = .974, RMSEA = .027 and PCLOSE = 0.998. Theoretically, the model confirms the predictive value of the selected variables, with respect to school failure. The results are applicable to both the design of educational policies and the direct intervention in the classroom. In both contexts, strategies can be developed that reduce factors that negatively affect school performance, actively linking students, teachers and parents.