The aim of this work is to examine potential of coactivation pattern analysis to find alterations in brain dynamics related to prodromal dementia with Lewy bodies (MCI-LB).We used resting state fMRI data (TR=0.98s, 580 scans) of 30 participants with MCI-LB (age 68.5 ± 6.0, 16 women) and 55 healthy controls (HC; age 67.7 ± 6.4, 33 women). The participants underwent neuropsychological examination and MRI measurement at Masaryk University in Brno (CZ). Inclusion in MCI-LB group was based on MCI-LB diagnostic criteria. Preprocessed fMRI data were analysed with Coactivation patten analysis. We analysed mean signal per network, originating in nodes of particular resting state network (frontoparietal control (FPCN), visual, default mode (DMN), dorsal attention and motor network). Differences between groups in brain dynamics (coverage - portion of particular pattern on total number of patterns) were tested with Wilcoxon rank sum test.We inspected optimal number of patterns analysing 2 to 40 coactivation patterns (CAPs). We identified 7 CAPs as optimum, based on elbow criterion applied on sum of squared errors. Coverage of coactivation pattern 4 showed statistically significant decrease in MCI-LB group (p=0.013) in comparison to HC group. CAP 4 and its coverages are depicted on Fig1.CAP 4 is characterised with high activation in DMN, moderate activation in visual network and negative activation in FPCN. This pattern probably represents mind wandering, occurred during the measurement. We plan to incorporate observed difference in classification model to test, if brain dynamics can serve as imaging biomarker for MCI-LB.