The current study aimed to acquire a better understanding of the relationships between metacognition, mathematics confidence, mathematics anxiety, general anxiety, gender and mathematics performance. The predictor variables investigated in this study were selected and simultaneously assessed as they have been identified in prior research to be related to mathematics performance (Morsanyi et al., 2019). However, previous research has predominantly only looked at these variables in isolation and the interaction between these predictors has only been investigated in a limited number of studies (Descender & Sansanguie, 2022; Lei et al., 2015). Here, 121 UK university students undertook four Likert-scale questionnaires to measure metacognition, mathematics confidence, mathematics anxiety and general anxiety, in addition to undertaking the CRT-Long (Primi et al., 2016) to measure mathematics performance. Results of a hierarchical multiple regression analysis indicated mathematics confidence independently predicted mathematics performance beyond the effect of mathematics anxiety, general anxiety, gender and metacognition for UK University students. Mathematics confidence also mediated the effect of mathematics anxiety on mathematics performance. Moreover, general anxiety was correlated to mathematics anxiety but did not explain additional variance in mathematics performance beyond the effect of the other predictors. A significant negative correlation between mathematics anxiety and mathematics confidence was also observed. The practical implications of these findings are centred around informing teaching practices within the education sector to shape the learning experience of anxious students by optimising learning environments and materials (Morsanyi et al., 2019; Plass et al., 2015). The findings from this investigation suggest that mathematics confidence training should be advocated and emphasised within classroom practice to improve mathematics performance.