OBJECTIVES: Impaired olfaction may be a biomarker for early Lewy body disease, but its value in mild cognitive impairment with Lewy bodies (MCI-LB) is unknown. We compared olfaction in MCI-LB with MCI due to Alzheimer's disease (MCI-AD) and healthy older adults. We hypothesized that olfactory function would be worse in probable MCI-LB than in both MCI-AD and healthy comparison subjects (HC). DESIGN: Cross-sectional study assessing olfaction using Sniffin' Sticks 16 (SS-16) in MCI-LB, MCI-AD, and HC with longitudinal follow-up. Differences were adjusted for age, and receiver operating characteristic (ROC) curves were used for discriminating MCI-LB from MCI-AD and HC. SETTING: Participants were recruited from Memory Services in the North East of England. PARTICIPANTS: Thirty-eight probable MCI-LB, 33 MCI-AD, 19 possible MCI-LB, and 32HC. MEASUREMENTS: Olfaction was assessed using SS-16 and a questionnaire. RESULTS: Participants with probable MCI-LB had worse olfaction than both MCI-AD (age-adjusted mean difference (B) = 2.05, 95% CI: 0.62-3.49, p = 0.005) and HC (B = 3.96, 95% CI: 2.51-5.40, p < 0.001). The previously identified cutoff score for the SS-16 of ≤ 10 had 84% sensitivity for probable MCI-LB (95% CI: 69-94%), but 30% specificity versus MCI-AD. ROC analysis found a lower cutoff of ≤ 7 was better (63% sensitivity for MCI-LB, with 73% specificity vs MCI-AD and 97% vs HC). Asking about olfactory impairments was not useful in identifying them. CONCLUSIONS: MCI-LB had worse olfaction than MCI-AD and normal aging. A lower cutoff score of ≤ 7 is required when using SS-16 in such patients. Olfactory testing may have value in identifying early LB disease in memory services.