Structural equation modelling was used to test a model integrating achievement goal orientation, learning style, self-efficacy and metacognition into a single framework that explained and predicted variation in performance. Self-efficacy was the strongest predictor of performance. Metacognition was a weak predictor of performance. Deep processing had a weak, negative relationship with performance. Mastery-approach goals were related to deep-processing and surface-processing learning styles. Mastery-approach and performance-approach goals were positive predictors of self-efficacy. Mastery-avoidance and performance-avoidance goals were related to self-efficacy.