Findings from the National Institute of Mental Health's Sequenced Treatment Alternatives to Relieve Depression trial indicate that approximately 50% of patients with major depressive disorder do not experience a treatment response after adequate first-line treatment with a selective serotonin reuptake inhibitor (SSRI). This study was designed to test the hypothesis that, after treatment failure with an SSRI, switching to venlafaxine extended release (ER) would offer advantages over switching to another SSRI, citalopram. The objectives of this trial were to compare the efficacy and safety of venlafaxine ER and citalopram in the treatment of moderate-to-severe depression in patients who did not experience a treatment response to an SSRI other than citalopram and to investigate the effects of severity of depression by categorizing treatment groups according to baseline severity. This was a 12-week, double-blind, randomized, parallel-group, multicenter study. Participants were adult outpatients who, following 8 weeks of monotherapy with an adequate dosing regimen of an SSRI other than citalopram and had not responded, met the diagnostic criteria for depression as described in the Diagnostic and statistical manual of mental disorders, fourth edition, and had a score ≥20 on the 21-item Hamilton Rating Scale for Depression (HAM-D21). After a 7-day screening period, patients were randomly assigned to receive venlafaxine ER 75 mg/day or citalopram 20 mg/day for the first 2 weeks. Doses could be increased every 2 weeks through week 6. Treatment lasted 12 weeks and was followed by a 1-week tapering period. Maximum dosages were venlafaxine ER 300 mg/day or citalopram 60 mg/day. The primary end point was the final on-therapy total HAM-D21 score. To investigate the treatment effects of the severity of depression on efficacy, a subgroup analysis was performed for baseline HAM-D21 total score ≤31 and >31. The analyses for HAM-D21, Montgomery–Åsberg Depression Rating Scale (MADRS), Clinical Global Impressions – Severity (CGI-S), and Clinical Global Impressions – Improvement scores were based on intent-to-treat (ITT) population, for both the primary analysis and subgroup analysis according to baseline HAM-D21 total scores ≤31 or >31. Safety assessments included the recording of adverse events (AEs). A total of 406 patients (200 venlafaxine ER, 206 citalopram) were randomly assigned and 396 patients were included in the ITT population (194 venlafaxine ER, 202 citalopram). Treatment groups were similar in terms of demographics and baseline psychiatric assessments. Two hundred and eighty-four patients (137 venlafaxine ER, 147 citalopram) were present in the ITT population with a baseline HAM-D21 total score ≤31 and 112 patients (57 venlafaxine ER, 55 citalopram) in the >31 group. In the primary analysis, there was no statistical difference between groups. The group with a baseline HAM-D21 total score of 20–31 did not differ significantly in any efficacy parameters. In the group with a baseline HAM-D21 total score >31, the venlafaxine ER group differed significantly from the citalopram group on the primary end point HAM-D21 total score (P=0.0121). The secondary end point CGI-S score was statistically significant (P=0.0359), although the MADRS total score (P=0.0930) was not. AEs were reported by 57.8 and 63.4% of venlafaxine ER and citalopram patients, respectively. Overall discontinuation rates were 24.5% for venlafaxine ER and 20.9% for citalopram. Discontinuation rates owing to an AE as a primary or secondary reason were 5.5% for venlafaxine ER and 5.3% for citalopram. Overall, venlafaxine ER and citalopram showed similar efficacy in patients who had an inadequate response to an SSRI. In the subset of more severely depressed patients, venlafaxine ER was significantly more effective on a number of efficacy measures. Patients who remain severely depressed following treatment with an SSRI may gain more benefit from the dual-action drug venlafaxine, rather than switching to another SSRI.