Generalised social phobia is a chronic disorder that commonly begins in childhood or early adolescence and tends to continue across the lifespan. Social phobia has one of the lowest remission rates of any mental disorder. It produces moderate to severe impact in people’s lives, affecting family, relationships, study, career, and medical problems. It is also commonly associated with other severe disorders such as depression and substance abuse. Empirically validated treatments for social phobia have been reasonably successful. The majority of empirically validated treatments centre around exposure in-vivo, often with additional cognitive restructuring. Several meta-analyses have reported that these comprehensive packages result in pre to post-treatment changes of around 0.8-1.0 sd on measures of social anxiety. Comparatively, social phobia has shown the smallest response to treatment of any anxiety disorder (Norton & Price, 2007). These findings suggest that improved techniques need to be developed for best practice management of social phobia. Arguably, these improvements should come from advances in theory. The current talk will describe three directions that we have attempted to take to increase effect sizes for the treatment of social phobia. The first direction comes from cognitive models of the maintenance of social phobia. These models have pointed to several factors that appear to play a central role in the maintenance of this disorder including a negative mental representation, subtle avoidance and safety strategies, and an inefficient focus of attention. Based on these suggestions it is predicted that incorporating strategies to target these mechanisms should improve effects. The results of a large-scale RCT will be described showing that a package that aims to modify these mechanisms produces a larger effect size change than a traditional treatment package utilizing only exposure and cognitive restructuring. A second direction of research picks up on recent discoveries in the area of cognitive bias modification. Experimental research has shown that modifying automatic attentional biases toward threat among anxious individuals can reduce fears. We are near to completing a RCT in which attention re-training has been incorporated into our best practice treatment package and the results of this trial will be reported. Finally, a new direction of research has shown that enhancement of the effects of exposure can be produced chemically by ingesting medications that directly target the biochemical mechanisms of learning. Incorporation of these methods into treatment may provide an enhanced exposure effect and may signal a new paradigm in the management of anxiety disorders.