There is urgent need for development and evaluation of targeted interventions for cognitive deficits in (partially) remitted major depression. Until now the analyses of the moderators of treatment efficacy were only examined in mixed samples of patients with schizophrenia, affective spectrum and schizoaffective disorders. Thus, the aim of our study was to evaluate the predictors of cognitive remediation therapy (CRT) improvement in a sample of (partially) remitted major depressive disorder patients.Reliable Change Index with corrections for practice effects was calculated for each participant as an indicator for training improvement. Thirty eight patients, who were randomized within our previously conducted CRT clinical trial, were divided into "Improvers" and "Nonimprovers" in the attention domain, to compare them on sociodemographic, psychopathological, neurocognitive, psychosocial and training factors.We detected 13 training participants who improved reliably in the attention domain. Illness duration was the only factor which significantly differentiated between Improvers and Nonimprovers. No significant differences between Improvers and Nonimprovers in terms of other clinical variables, sociodemographic and neuropsychological factors were found.Exploratory research results should be taken with caution. Focus on the attention domain could have led to a limited point of view.Our findings represent a first analysis of the predictors of cognitive remediation training improvement in (partially) remitted unipolar depression. Much more work should be done to refine cognitive treatment approaches. An initiation of cognitive training in early stages of the disease could be beneficial for the affected patients.