Behaviour therapy consisting of specific exposure in vivo is reviewed in terms of its clinical usefulness for the treatment of agoraphobia, obsessive compulsive neurosis and related conditions. A combined behavioural and psychological formulation of individual target problems is advocated. These can be rated in terms of discomfort, disability, and physiological indices of anxiety. Although discomfort and disability have been found to correlate, their relationship with anxiety measures is questioned. Variables affecting the outcome of exposure are reviewed, including: duration--best results with prolonged, frequent sessions over a short period; arousal--experience of high anxiety not essential; adjunctive cognitive therapy--considered to add nothing to effectiveness during exposure treatment. The author goes on to argue that simple response prevention is effective in reducing ritualising activities in obsessive compulsive neurosis, without effecting concomitant anxiety. Whereas anxiety responds independently to exposure therapy, and hence that the anxiety reduction model of obsessive compulsive neurosis is inadequate. The putative anti-obsessional properties of clomipramine are questioned, and the utility of imipramine in agoraphobia is disputed.