The contrastive analysis is made based on the data from automatic weather stations (AWS) and conventional stations from January 2003 to December 2004 at Yanzhou Weather Station. All the essential meteorological elements are compared and analyzed, and the reasons for discrepancy are discussed. The result shows that though there is certain discrepancy in data from two observation approaches, the errors are within the permitted limits. The discrepancies in average ground minimum temperature, average ground maximum temperature, ground extreme minimum temperature, and ground extreme maximum temperature are relatively greater. The causes of the discrepancies (except for those resulted from the design and capability of the instruments), the changes of observation environment, and factors relevant to artificial operating and special weather phenomenon are also discussed. The findings are of significance to the application studies of automatic weather stations and the precision of AWS observation data.