The relationship between anxiety and interpretive bias has been studied extensively, but the causal direction of this relationship remains largely unexplored. Do negative interpretations cause anxiety or is anxiety the cause of negative interpretations? Or are the two mutually reinforcing? The present study addressed this issue by experimentally inducing either a negative or a positive interpretive bias using Mathews and Mackintosh' [(2002). Induced emotional interpretation bias and anxiety. Journal of Abnormal Psychology, 109, 604-615] training paradigm and then examining its impact on state anxiety and anxiety vulnerability. In addition, it was investigated as to whether the interpretive bias was trained implicitly. Results indicated that style of interpreting could be manipulated. That is, when confronted with ambiguous information after the training, participants (n=118) interpreted this information congruent with their (positive or negative) training condition. Data on the issue of implicitness showed that participants tended to be explicitly aware of the valence of their training stimuli. Effects of trained interpretive bias on anxiety were only marginal and absent on anxiety vulnerability. It appears that interpretive bias can be trained reliably, but its effects on mood and vulnerability require further explanation.