This thesis investigated negative attitudes towards mathematics and, in particular, mathematics anxiety. Studies 1 and 2 were based on first year university student samples while the subjects in Study 3 were high school pupils. These studies obtained data about levels of mathematics anxiety, the factor structure of a mathematics anxiety scale, the Revised Mathematics Anxiety Rating Scale (RMARS), and the relationship among mathematics anxiety, other anxiety measures and the cognitions of subjects before and after a mathematics task. It was found that there were two major dimensions to mathematics anxiety; anxiety about learning mathematics and anxiety about mathematics tests. Scores on the RMARS were not high in any of the three samples. Multivariate analyses indicated that both affective reactions and cognitions, which were obtained by thought-listing and cognitive checklist, contributed to mathematics anxiety, with cognitions being particularly important determinants for females. The first three studies suggested that students' concerns about mathematics may involve negative feelings other than anxiety. Accordingly, Studies 4 and 5 took a broader perspective and examined more generalized negative mathematics attitudes. The Repertory Grid Technique was used in both of these later studies and yielded individual profiles of attitudes from both university and high school students. An important finding from the five studies was that negative mathematics attitudes, including mathematics anxiety, consisted of both affective and cognitive components. It was concluded that students with negative mathematics attitudes may not necessarily be mathematics anxious. It was suggested that it may be desirable to assess negative mathematics attitudes on an individual basis, for which purpose the Repertory Grid is a suitable instrument.