Based　on　53　years　observed　data　of　rainfall　in　Beijing,　the　annual　maximum　series　method,　the　annual　exceedance　series　method,　the　partial　duration　series　(rainfall　depth　over　threshold)　method　and　the　annual　multi-sampling　series　method　were　separately　used　to　process　rainfall　depth　data.　Four　candidate　distributions　(log　normal,　Pearson　III,　log　Pearson　III　and　Gumbel)　were　used　to　fit　the　observed　data　and　find　the　most　appropriate　distribution.　Distribution　parameters　were　estimated　using　the　moment　method.　The　assessment　was　based　on　the　Chi-Squared　test　and　Kolmogorov-Smirnov　test　combined　with　root　mean　squared　error.　On　the　basis　of　these　comparisons　and　estimation　results,　it　can　be　concluded　that,　if　the　length　of　rainfall　records　is　shorter　or　selected　return　period　is　less　than　one　year,　the　annual　multi-sampling　series　method　is　proposed,　otherwise,　the　annual　maximum　series　method　is　proposed.　The　Pearson　III　distribution　is　the　most　appropriate　distribution　for　Beijing　rainfall　data.