Background Transcranial direct current stimulation (tDCS) is a non-invasive form of neurostimulation with potential for development as a self-administered intervention. It has shown promise as a safe and effective treatment for obsessive compulsive disorder (OCD) in a small number of studies. The two most favourable stimulation targets appear to be the left orbitofrontal cortex (L-OFC) and the supplementary motor area (SMA). We report the first study to test these targets head-to-head within a randomised sham-controlled trial. Our aim was to inform the design of future clinical research studies, by focussing on the acceptability and safety of the intervention, feasibility of recruitment, adherence to and tolerability of tDCS, and the size of any treatment-effect. Methods FEATSOCS was a randomised, double-blind, sham-controlled, cross-over, multicentre study. Twenty adults with DSM-5-defined OCD were randomised to treatment, comprising three courses of clinic-based tDCS (SMA, L-OFC, Sham), randomly allocated and delivered in counterbalanced order. Each course comprised four 20-min 2 mA stimulations, delivered over two consecutive days, separated by a ‘washout’ period of at least four weeks. Assessments were carried out by raters who were blind to stimulation-type. Clinical outcomes were assessed before, during, and up to four weeks after stimulation. Patient representatives with lived experience of OCD were actively involved at all stages. Results Clinicians showed willingness to recruit participants and recruitment to target was achieved. Adherence to treatment and study interventions was generally good, with only two dropouts. There were no serious adverse events, and adverse effects which did occur were transient and mostly mild in intensity. Yale-Brown Obsessive-Compulsive Scale (Y-BOCS) scores were numerically improved from baseline to 24 h after the final stimulation across all intervention groups but tended to worsen thereafter. The greatest effect size was seen in the L-OFC arm, (Cohen’s d = −0.5 [95% CI −1.2 to 0.2] versus Sham), suggesting this stimulation site should be pursued in further studies. Additional significant sham referenced improvements in secondary outcomes occurred in the L-OFC arm, and to a lesser extent with SMA stimulation. Conclusions tDCS was acceptable, practicable to apply, well-tolerated and appears a promising potential treatment for OCD. The L-OFC represents the most promising target based on clinical changes, though the effects on OCD symptoms were not statistically significant compared to sham. SMA stimulation showed lesser signs of promise. Further investigation of tDCS in OCD is warranted, to determine the optimal stimulation protocol (current, frequency, duration), longer-term effectiveness and brain-based mechanisms of effect. If efficacy is substantiated, consideration of home-based approaches represents a rational next step. Trial registration: ISRCTN17937049. https://doi.org/10.1186/ISRCTN17937049