Abstract Background Persecutory delusions are the most common type of delusions in psychosis and present in around 10–15% of the general population. Persecutory delusions are thought to be sustained by biased cognitive and emotional processes. Recent advances favour targeted interventions, focussing on specific symptoms or mechanisms. Our aim is to test the clinical feasibility of a novel psychological intervention, which manipulates biased interpretations toward more adaptive processing, in order to reduce paranoia in patients. Methods The ‘Cognitive Bias Modification for paranoia’ (CBM-pa) study is a feasibility, double-blind, randomised controlled trial (RCT) for 60 stabilised outpatients with persistent, distressing paranoid symptoms. Patients will be randomised at a 50:50 ratio, to computerised CBM-pa or a text-reading control intervention, receiving one 40-min session per week, for 6 weeks. CBM-pa involves participants reading stories on a computer screen, completing missing words and answering questions about each story in a way that encourages more helpful beliefs about themselves and others. Treatment as Usual will continue for patients in both groups. Patients will be assessed by a researcher blind to allocation, at baseline, each interim session, post treatment and 1- and 3-month follow-up post treatment. The primary outcome is the feasibility parameters (trial design, recruitment rate and acceptability) of the intervention. The secondary outcomes are clinical symptoms (including severity of paranoia) as assessed by a clinical psychologist, and ‘on-line’ measurement of interpretation bias and stress/distress. The trial is funded by the NHS National Institute for Health Research. Discussion This pilot study will test whether CBM-pa has the potential to be a cost-effective, accessible and flexible treatment. If the trial proves feasible and demonstrates preliminary evidence of efficacy, a fully powered RCT will be warranted. Trial registration Current Controlled Trials ISRCTN: 90749868 . Retrospectively registered on 12 May 2016.