The current society is based on science and technology, depending partly on mathematics. It leads to citizens’ success in school mathematics, being measured through achievement, which can be predicted by affective, cognitive, and behavioral variables. The aim of this study was to determine the extent to which self-concept, learning strategies, attitude towards science and mathematics, school environment, and previous scores in science and mathematics predict achievement in mathematics. A convenience sample of 352 pupils taking part in a science, technology, engineering, and mathematics (STEM) experimentation outreach program belonging to state schools and state-funded schools from rural and urban environments was analysed. The instrument was composed of the Auzmendi scale of attitude towards mathematics modified, the attitude towards school science, the AUDIM questionnaire for self-concept (physical, social, personal, academic, and general), and the CEA questionnaire for learning strategies (emotional control, critical and creative thinking, and metacognition). Sex, type of school, and school environment were covariates. A binary logistic regression model was obtained for mathematics achievement, which correctly classified 82.1% of students, with previous science and mathematics achievement, science achievement, and critical and creative thinking as predictors, and urban schools playing a positive role. Implications of these predictors on mathematics education are discussed.