Cognitive-behavioural therapy (CBT) has a wide-ranging empirical base, supporting its place as the evidence-based treatment of choice for the majority of psychological disorders. However, many clinicians feel that it is not appropriate for their patients, and that it is not effective in real life-settings (despite evidence to the contrary). This paper addresses the contribution that we as clinicians make to CBT going wrong. It considers the evidence that we are poor at implementing the full range of tasks that are necessary for CBT to be effective--particularly behavioural change. Therapist drift is a common phenomenon, and usually involves a shift from 'doing therapies' to 'talking therapies'. It is argued that the reason for this drift away from key tasks centres on our cognitive distortions, emotional reactions, and use of safety behaviours. A series of cases is outlined in order to identify common errors in clinical practice that impede CBT (and that can make the patient worse, rather than better). The principles behind each case are considered, along with potential solutions that can get us re-focused on the key tasks of CBT.