Abstract: Objective To analyze the effect of geographically stratified early detection of communicable disease outbreak and provide evidence for the further improvement of the warning method. Methods Communicable disease real-time surveillance and early system was developed by using Microsoft Excel software and Visual Basic for Application program. The early warning period was 7 days, the early warning threshold was set according to the national definitions of public health emergency and outbreak. The report cards were downloaded from the communicable disease reporting system and used in the early warning system. Three surveillance levels were set, i.e. county level, township level and village/school level. The actual case number was compared with early warning threshold respectively, and finally the village or school where cases clustered was determined. Thus early warning information was obtained and related response was made. The analyses on the reported case number, early warning signal number, and verifying or field survey results were condcuted. The evaluation indicators included the response rate of early warning signals, proportion of suspect events, field survey rate, sensitivity and positive predictive value. Results From 2007 to 2013, a total of 148 early warning signals, involving 14 communicable diseases, were obtained through geographically stratified early detection of communicable disease outbreak. After verifying, 127 suspect events (85.81%), involving 10 communicable diseases, were confirmed. The response rate within 2 hours was 100.00%. After field survey, 85 outbreaks were confirmed. The survey rate within 24 hours was 100.00%. The sensitivity was 100.00% and the positive predictive value was 57.43%. Conclusion This early waning method is feasible and accurate, which can be used in the detection of communicable outbreaks in early stage, but further improvement is still needed.