Integrating accessibility by public transit with land use planning is a crucial precondition for sustainable urban development. Accessibility by public transit has been widely assessed in a GIS environment using aggregated zonal data, such as traffic analysis zones, census tracts, dissemination areas, dissemination blocks, 200 * 200 m grids and 50 * 50 m grids. Nevertheless, it has been proved that the scale and zoning scheme of zones may alter analysis results, which is known as the Modifiable Areal Unit Problem (MAUP). Therefore, it is essential to know how the MAUP affects assessment of accessibility. This research addressed the MAUP effects, when evaluating accessibility based on cumulative opportunity measures. This research applied a cumulative accessibility measure, which calculated accessibility in terms of the number of urban nodes that could be reached within a given travel time or distance. The City of Windsor, Canada, was used as the study area. The MAUP effects were examined based on 6 types of zones (e.g. census tracts, dissemination areas, dissemination blocks, 0.6 km, 0.3 km and 0.15 km grids) at comparable scales or zoning schemes. It was found that the MAUP may significantly alter assessment results of accessibility and should be paid highly attention to. The two outcomes of the MAUP effects on accessibility measurements are: changes of accessibility score and alterations of policy implications that are based on accessibility measurements. Three ways were discussed to deal with the MAUP impacts on accessibility measurements: using disaggregate data if possible, using low aggregated data and selecting zones according to research purposes.