BACKGROUND AND OBJECTIVES: It is well established that attention bias and interpretation bias each have a key role in the development and continuation of anxiety. How the biases may interact with one another in anxiety is, however, poorly understood. Using cognitive bias modification techniques, the present study examined whether training a more positive interpretation bias or attention bias resulted in transfer of effects to the untrained cognitive domain. Differences in anxiety reactivity to a real-world stressor were also assessed. METHODS: Ninety-seven first year undergraduates who had self-reported anxiety were allocated to one of four groups: attention bias training (n = 24), interpretation bias training (n = 26), control task training (n = 25) and no training (n = 22). Training was computer-based and comprised eight sessions over four weeks. Baseline and follow-up measures of attention and interpretation bias, anxiety and depression were taken. RESULTS: A significant reduction in threat-related attention bias and an increase in positive interpretation bias occurred in the attention bias training group. The interpretation bias training group did not exhibit a significant change in attention bias, only interpretation bias. The effect of attention bias training on interpretation bias was significant as compared with the two control groups. There were no effects on self-report measures. LIMITATIONS: The extent to which interpretive training can modify attentional processing remains unclear. CONCLUSIONS: Findings support the idea that attentional training might have broad cognitive consequences, impacting downstream on interpretive bias. However, they do not fully support a common mechanism hypothesis, as interpretive training did not impact on attentional bias.