Our study aimed to analyse the olfactory dysfunction (OD) evaluations between self-report, the Hyposmia Rating Scale (HRS) and the Sniffin' Sticks test, and the relationship between OD and clinical features of AD. Sixty patients with AD dementia, 37 patients with mild cognitive impairment (MCI) due to AD and 30 healthy controls were consecutively recruited. Olfactory function was evaluated by self-report, HRS and Sniffin' Sticks test. Patients were divided into AD with OD (AD-OD) and AD with no OD (AD-NOD) groups based on the results of the Sniffin' Sticks test. Cognitive symptoms and neuropsychiatric symptoms were assessed by corresponding scales, and activities of daily living (ADL) were assessed by the ADL scale. In the control, MCI due to AD and AD dementia groups, the frequency of OD was 10.0%, 13.5% and 18.3%, respectively, by self-report; 6.7%, 24.3% and 48.3%, respectively, by HRS; and 3.3%, 13.5% and 65.0%, respectively, by the Sniffin' Sticks test. Compared to the results of the Sniffin' Sticks test, the diagnostic coincidence rates of OD by HRS in patients with MCI due to AD and AD dementia were 89.2% and 66.7%, respectively. Compared to the AD-NOD group, the scores of global cognition and memory, visuospatial ability and attention were all decreased (P<0.05), the apathy score was increased (P<0.05), and the ADL score was elevated (P<0.01). The frequency and accuracy of OD by self-report is relatively low. HRS can be used for screening olfaction in patients with MCI due to AD. The Sniffin' Sticks test can be used for validating OD in AD patients. AD-OD patients have severe impairments in global cognition and multiple cognitive domains of memory, visuospatial ability and attention, as well as neuropsychiatric symptoms of apathy, and thus have seriously compromised ADL.