BackgroundComputerized cognitive bias modification for social anxiety disorder has in several well conducted trials shown great promise with as many as 72% no longer fulfilling diagnostic criteria after a 4 week training program. To test if the same program can be transferred from a clinical setting to an internet delivered home based treatment the authors conducted a randomized, double-blind placebo-controlled trial.MethodsAfter a diagnostic interview 79 participants were randomized to one of two attention training programs using a probe detection task. In the active condition the participant was trained to direct attention away from threat, whereas in the placebo condition the probe appeared with equal frequency in the position of the threatening and neutral faces.ResultsResults were analyzed on an intention-to-treat basis, including all randomized participants. Immediate and 4-month follow-up results revealed a significant time effect on all measured dimensions (social anxiety scales, general anxiety and depression levels, quality of life). However, there were no time x group interactions. The lack of differences in the two groups was also mirrored by the infinitesimal between group effect size both at post test and at 4-month follow-up.ConclusionWe conclude that computerized attention bias modification may need to be altered before dissemination for the Internet.Trial registrationISRCTN01715124