BACKGROUND: The Diagnostic and Statistical Manual fifth edition (DSM-5) provides new criteria for delirium diagnosis. We examined delirium diagnosis using these new criteria compared with the Diagnostic and Statistical Manual fourth edition (DSM-IV) in a large dataset of patients assessed for delirium and related presentations. METHODS: Patient data (n = 768) from six prospectively collected cohorts, clinically assessed using DSM-IV and the Delirium Rating Scale-Revised-98 (DRS-R98), were pooled. Post hoc application of DRS-R98 item scores were used to rate DSM-5 criteria. 'Strict' and 'relaxed' DSM-5 criteria to ascertain delirium were compared to rates determined by DSM-IV. RESULTS: Using DSM-IV by clinical assessment, delirium was found in 510/768 patients (66%). Strict DSM-5 criteria categorized 158 as delirious including 155 (30%) with DSM-IV delirium, whereas relaxed DSM-5 criteria identified 466 as delirious, including 455 (89%) diagnosed by DSM-IV (P <0.001). The concordance between the different diagnostic methods was: 53% (ĸ = 0.22) between DSM-IV and the strict DSM-5, 91% (ĸ = 0.82) between the DSM-IV and relaxed DSM-5 criteria and 60% (ĸ = 0.29) between the strict versus relaxed DSM-5 criteria. Only 155 cases were identified as delirium by all three approaches. The 55 (11%) patients with DSM-IV delirium who were not rated as delirious by relaxed criteria had lower mean DRS-R98 total scores than those rated as delirious (13.7 ± 3.9 versus 23.7 ± 6.0; P <0.001). Conversely, mean DRS-R98 score (21.1 ± 6.4) for the 70% not rated as delirious by strict DSM-5 criteria was consistent with suggested cutoff scores for full syndromal delirium. Only 11 cases met DSM-5 criteria that were not deemed to have DSM-IV delirium. CONCLUSIONS: The concordance between DSM-IV and the new DSM-5 delirium criteria varies considerably depending on the interpretation of criteria. Overly-strict adherence for some new text details in DSM-5 criteria would reduce the number of delirium cases diagnosed; however, a more 'relaxed' approach renders DSM-5 criteria comparable to DSM-IV with minimal impact on their actual application and is thus recommended.