Depression has been widely associated with a cognitive deficit leading to the negative interpretation of ambiguous information. Recently, cognitive bias modification (CBM) procedures have shown that such negative biases are causally related to emotional vulnerability. However, research using CBM has been notably lacking in depression. This is the first double blind randomised controlled study investigating the effect of cognitive bias modification-errors (CBM-errors), on depression and its influence on mood and resilience to stress. CBM-errors is a new form of cognitive bias modification for interpretation, which targets the full range of cognitive errors, as well as interpretation biases. Forty clinically depressed participants were randomly allocated to a positive training group or neutral text reading control group. Participants trained to make positive interpretations subsequently interpreted novel ambiguous information in a positive manner compared to controls. The results suggest that a positive cognitive bias can be induced in clinically depressed individuals using a simple computerised intervention. There was little evidence of corresponding benefits in terms of mood or response to stress, suggesting that multiple sessions are likely to be needed to confer symptom related change. A systematic investigation of the optimum number and timing of multiple sessions is now called for.