This article describes a multidisciplinary approach to assessing potential vegetation types. The relation between vegetation distribution as derived from field survey and habitat characteristics in the Yellow River Delta (YRD) was analyzed using detrended canonical correspondence analysis (DCCA). The three-dimensional groundwater numerical simulation model MODFLOW yielded an accurate distribution of the shallow groundwater table within the study area. Groundwater table and soil salinity appeared to have the largest influence on the distribution patterns of specific indicator species. The quantitative relationships between species presence and environmental factors were further explored using logistic regression, allowing for the prediction of potential species distribution in relation to fluctuations in groundwater depth and soil salinity. Based on this, potential distribution maps for specific plant species and their communities were generated using ArcGIS. The potential vegetation map was compared to the actual vegetation map (interpreted from SPOT imagery), which leads to a discussion of the main factors responsible for fragmentation and degradation of the local vegetation system. Copyright © 2010 John Wiley & Sons, Ltd.