The Mini-Mental State Examination (MMSE) and the Montreal Cognitive Assessment (MoCA) are the most commonly used scales to test cognitive impairment in Lewy body disease (LBD), but there is no consensus on which is best suited to assess cognition in clinical practice and most sensitive to cognitive decline. Retrospective cohort study of 265 LBD patients [Parkinson’s disease (PD) without dementia (PDnD, N = 197), PD with dementia (PDD, N = 40), and dementia with Lewy bodies (DLB, N = 28)] from an international consortium who completed both the MMSE and MoCA at baseline and 1-year follow-up (N = 153). Percentage of relative standard deviation (RSD%) at baseline was the measure of inter-individual variance, and estimation of change (Cohen’s d) over time was calculated. RSD% for the MoCA (21 %) was greater than for the MMSE (13 %) (p = 0.03) in the whole group. This difference was significant only in PDnD (11 vs. 5 %, p < 0.01), but not in PDD (30 vs. 19 %, p = 0.37) or DLB (15 vs. 14 %, p = 0.78). In contrast, the 1-year estimation of change did not differ between the two tests in any of the groups (Cohen’s effect <0.20 in each group). MMSE and MoCA are equal in measuring the rate of cognitive changes over time in LBD. However, in PDnD, the MoCA is a better measure of cognitive status as it lacks both ceiling and floor effects.