In order to study the characteristics of regional drought and flood and its impact on crop yield, this paper discusses the evolution law of drought and flood in each growth stage of crop and the impact of drought and flood in different growth stages on crop yield using the technology of information diffusion and chaotic time series analysis based on the comparison of SPEI and precipitation temperature homogenization index. With Xi'an as a typical region and corn as the research object, we analyzed the meteorological data and corn yield data from 1951 to 2015. The results show that: firstly, the trend of drought and flood in each growth stage of corn in Xi'an with time shows the form of wave crest and wavetrough alternately and a certain periodicity. Each growth stage of corn shows a trend of relative abundant water after 2008. Secondly, most of the growth stages of corn are in normal state, and the frequency is about 45%. The frequency maximum of severe drought is 4.62%, which occurred in emergence-jointing, and the frequency maximum of above moderate wetness is 15.39%, which occurred in the jointing-tasselling phase. Thirdly, the largest Lyapunov exponents of all growth phases are greater than 0. Time series of drought and flood indicate chaotic feature to a certain extent. Finally, the jointing-tasselling stage is compared with other growth stages, and the correlation between yield and drought and flood is obvious, which reflects the sensitivity of this stage to water to some extent. This research is of guiding implications to disaster prevention and yield reduction of corn in Xi'an city.