Ardeidae and their habitat are being threatened by the destruction of wetland ecosystems. Because the availability of suitable Ardeidae habitats are gradually decreasing, habitat suitability evaluations using remote sensing and measurement technology are becoming research hotspots. In this study, Mai Po and the Deep Bay Area in Hong Kong was chosen as a case study area. Based on descriptive waterfowl statistics from field observation data, we used Moran's index to assess spatial autocorrelation. The results of this analysis indicated that the field data was randomly distributed within the study area, and that there were no duplicate records. Therefore the data was suitable for modeling. We used a logistic regression model to determine the factors associated with the waterfowls' habitat. First, according to Ardeidae habitat dependencies, we collected data of 15 factors relevant to the waterfowl in January 2003 using Geographic Information System (GIS) and Remote Sensing technology. We randomly selected 1000 points from each factor's grid map and used the leverage value and cook distance method to look up unusual points. Few unusual points were identified, and most of the selected points were able to be used to establish the binary logistic regression model. Nine factors were identified by the model as having an important influence on the presence/absence of Ardeidae. These factors were land use, Normalized Difference Vegetation Index, slope, rainfall, TM 4 texture, TM 3 texture, road density, road distance, and human habitation density. The Nagelkerke R² verified that the model was a good fit (coefficient =0.774). Aspect, digital elevation, human habitation distance, TM 2 texture, relative humidity, and temperature were simultaneously filtered out of the model, suggesting that the waterfowls' habitat preferences were not influenced by these factors. We then used a quick clustering method to grade habitat suitability, and found that the case area could be divided into five levels. From a two figure stack in GIS, we selected 500 points from the Ardeidae population. We found that the prediction data from the binary logistic regression model was in good agreement with the observed field data of Ardeidae waterfowl in this area, with a kappa coefficient of 0.774. We suggest that the suitability grades basically correspond with the waterfowls' habitat, and that we could predict the presence of waterfowl with the model. The model could also be used to estimate the presence of waterfowl in unstudied areas that are difficult to access. The model has better application prospects in studies of wetlands. Finally, we used January, 2009 data of the nine significant factors in Mai Po and the Deep Bay Area to test the universality of the model's equation. We followed the same steps as for the 2003 data, and the fitting accuracy reached a kappa coefficient of 0.758. The results of our study show that the binary logistic regression model could be used for forecasting Ardeidae waterfowl habitat suitability. We demonstrate the effectiveness of correlation analyses for predicting waterfowl habitat. Our model performed well in the prediction of the presence of Ardeidae waterfowl, and provides a reference for the protection and management of waterbird habitats, and for future studies.