The aim was to identify whether performance on olfactory identification can distinguish neurological/neurodegenerative disorders (NNDs) from primary psychiatric disorders (PPDs). This is a cross-sectional retrospective study of inpatients assessed in Neuropsychiatry, Royal Melbourne Hospital. Data extracted from the admission records included: demographics, tobacco use, medical comorbidities, cognitive function using the Neuropsychiatry Unit Cognitive Assessment Tool (NUCOG), and odor identification using the Sniffin' Sticks Screening 12 test. The final diagnosis for patients was informed by established diagnostic criteria. A total 121 patients were included. Eighty-eight patients (73%) were diagnosed with neurological or neurodegenerative disease, including Alzheimers dementia, frontotemporal dementia, Lewy body parkinsonian-related dementias (Parkinson disease, multiple system atrophy, dementia with Lewy bodies) and other neurological causes of dementia; 33 patients (27%) were diagnosed with PPDs (including mood and psychotic disorders). Patients who scored ≤8 on the Sniffin' Sticks Screening 12 test were more likely to have NND than PPD, even after adjustment for age, sex and tobacco use (P=0.009, adjusted odds ratios=3.85, 95% confidence interval=1.40-10.62). Receiver operating characteristic curve analyses demonstrated that a score of ≤8 differentiated NND from PPD with sensitivity of 57% and specificity of 73% (receiver operating characteristic area under the curve of 0.67, P=0.004). Patients with neuropsychiatric difficulties who score 8 or less on Sniffin' Sticks are more likely to have a neurodegenerative illness. A cut-off score of 8 is potentially a "red flag" for clinicians faced with the diagnostic question of PPD versus NND.