Electroconvulsive therapy (ECT) is one of the most effective and rapid-acting treatment for severe depression but is associated with cognitive side-effects. Identification of add-on treatments that counteract these side-effects would be very helpful. This randomized, double-blinded, placebo-controlled, parallel-group study investigated the effects of four add-on erythropoietin (EPO; 40,000 IU/ml) or saline (placebo) infusions over 2.5 weeks of ECT (eight ECT sessions) in severely depressed patients with unipolar or bipolar depression. Neuropsychological assessments were conducted pre-ECT, three days after the eighth ECT (week 4), and at a 3-month follow-up. Further, functional magnetic resonance imaging (fMRI) was conducted after the eighth ECT. The primary outcome was change from pre- to post-ECT in a ‘speed of complex cognitive processing’ composite. Secondary outcomes were verbal and autobiographical memory. Of sixty randomized patients, one dropped out before baseline. Data were thus analysed for 59 patients (EPO, n = 33; saline, n = 26), of whom 28 had fMRI data. No ECT-related decline occurred in the primary global cognition measure (ps≥0.1), and no effect of EPO versus saline was observed on this outcome (ps≥0.3). However post-ECT, EPO-treated patients exhibited faster autobiographical memory recall than saline-treated patients (p = 0.02), which was accompanied by lower memory-related parietal cortex activity. The absence of global cognition changes with ECT and EPO, coupled with the specific impact of EPO on autobiographical memory recall speed and memory-related parietal cortex activity, suggests that assessing autobiographical memory may provide increased sensitivity in evaluating and potentially preventing cognitive side-effects of ECT.