This study examines the structural relationship among motivation, deep learning approach, and academic achievement of middle school students in Turkey. Participants were 746 seventh grade and eighth grade students enrolled in public middle schools in Sinop and Ankara, Turkey. Motivated Strategies for Learning Questionnaire, Study Process Questionnaire (R-SPQ-2F) and GPA scores of participants were used in the study. Data were analyzed by Structural Equation Modeling. The results of the study revealed that motivational variables are related with the use of deep learning approach which is related with higher GPA. Path analyses demonstrated that deep learning approach fully mediated the relationship between students' motivational variables and academic achievement. Self-efficacy, task value, and intrinsic goal orientation as the indirect effects through deep learning approach on academic achievement were strong predictors in the model. [This study was presented at the International Congress of Eurasian Educational Researches held in Ankara on June 19-22, 2019.]