select agentid,count(id) as allnum, sum(if(review_flag = 'p'and delflag in ('n' , 'p'),1,0)) as approved_1, sum(if(review_flag = 'e', 1, 0)) as approved_2, sum(if(review_mode = '3' and review_flag = 'y',1,0)) as approved_3, sum(if(review_mode = '2' and review_flag = 'y',1,0)) as approved_4, sum(if(review_mode = '1' and review_flag = 'y',1,0)) as approved_5, sum(if(review_flag = 'n' and resubmit = 'y',1,0)) as approved_6, sum(if(review_flag = 'n' and resubmit = 'n',1,0)) as approved_7 from talbe where agentid in ('a' , 'b', 'c') and delflag in ('n' , 'p', 'h') group by agentid 该语句执行的时候 需要20多s才完成。

回复内容： select agentid,count(id) as allnum, sum(if(review_flag = 'p'and delflag in ('n' , 'p'),1,0)) as approved_1, sum(if(review_flag = 'e', 1, 0)) as approved_2, sum(if(review_mode = '3' and review_flag = 'y',1,0)) as approved_3, sum(if(review_mode = '2' and review_flag = 'y',1,0)) as approved_4, sum(if(review_mode = '1' and review_flag = 'y',1,0)) as approved_5, sum(if(review_flag = 'n' and resubmit = 'y',1,0)) as approved_6, sum(if(review_flag = 'n' and resubmit = 'n',1,0)) as approved_7 from talbe where agentid in ('a' , 'b', 'c') and delflag in ('n' , 'p', 'h') group by agentid 该语句执行的时候 需要20多s才完成。

调整表结构. 似乎你把很复杂的逻辑放在 数据表结构里了.
多次sum不是问题. 主要在于这个group by上, 一般的, mysql 全表扫一边, 创建临时表, 排序分组. 这个是很慢的. 你可以试一下 建立 (agentid, delflag, review_flag, review_mode, resubmit, id) 的联合索引.
索引，从新设计你自己标的结构。。。怎么会sum这么多次。。。
这不是sum的问题啊，这条语句本来就复杂。
这种情况我想应该是索引建立的不够科学吧，你可以借助explain来具体查看什么地方会卡住，是否用到索引。
同时，给你一个建议，你应该把表结构贴出来，这样路过的大神就能具体帮你分析分析了。
先分组 sum，把结果取到你的程序里去继续加工