This is a secondary data analysis from our erythropoietin (EPO) trials. We examine (I) whether EPO improves speed of complex cognitive processing across bipolar and unipolar disorder, (II) if objective and subjective baseline cognitive impairment increases patients׳ chances of treatment-efficacy and (III) if cognitive improvement correlates with better subjective cognitive function, quality of life and socio-occupational capacity. Patients with unipolar or bipolar disorder were randomized to eight weekly EPO (N=40) or saline (N=39) infusions. Cognition, mood, quality of life and socio-occupational capacity were assessed at baseline (week 1), after treatment completion (week 9) and at follow-up (week 14). We used repeated measures analysis of covariance to investigate the effect of EPO on speed of complex cognitive processing. With logistic regression, we examined whether baseline cognitive impairment predicted treatment-efficacy. Pearson correlations were used to assess associations between objective and subjective cognition, quality of life and socio-occupational capacity. EPO improved speed of complex cognitive processing across affective disorders at weeks 9 and 14 (p≤0.05). In EPO-treated patients, baseline cognitive impairment increased the odds of treatment-efficacy on cognition at weeks 9 and 14 by a factor 9.7 (95% CI:1.2-81.1) and 9.9 (95% CI:1.1-88.4), respectively (p≤0.04). Subjective cognitive complaints did not affect chances of treatment-efficacy (p≥0.45). EPO-associated cognitive improvement correlated with reduced cognitive complaints but not with quality of life or socio-occupational function. As the analyses were performed post-hoc, findings are only hypothesis-generating. In conclusion, pro-cognitive effects of EPO occurred across affective disorders. Neuropsychological screening for cognitive dysfunction may be warranted in future cognition trials.