Using the parallel observation data of 135 stations with AWS (Automatic Weather Station) and manual pressure measurements in Sichuan, the differences between the two sorts of data are analyzed and discussed. The results show that the daily difference does not satisfy the normal distribution pattern, and the average daily pressure observed by manual way is 0.35 hPa greater than that by AWS. The correlation coefficient of average daily pressure is 099 with the significant level being 0.01； about 3/4 of manual measurements are greater than those from AWS’s. The monthly average difference between automatic and manual measurements is larger in the first half year with the largest difference in June. The annual mean difference is smallest in 2004 and greatest in 2006. There are about 90% stations in the whole province with negative difference values.