更通用的解决方案是：

select e.*, t2.*
  from (
    select event,
      (select coalesce(sum(chance_percentage), 0) 
         from table1 t2 where t2.event < t1.event) as lower_bound,
      (select sum(chance_percentage) 
         from table1 t3 where t3.event <= t1.event) as upper_bound
      from table1 t1) e
  join (select 100.0 * rand() as p) t2
    where t2.p >= e.lower_bound and t2.p < e.upper_bound;

做一个累加和然后运行 rand() 一次：

select t.event
from (select t.*, (@cume_p = @cume_p + p) as cume_p
      from t cross join
           (select @cume_p := 0, @rand = rand()) params
     ) t
where @rand >= cume_p - p and
      @rand < cume_p;

请注意 rand() 只叫过一次。该值存储在变量中；这是一个任意的选择。也可以在子查询中：

select t.event
from (select t.*, (@cume_p = @cume_p + p) as cume_p
      from t cross join
           (select @cume_p := 0) params
     ) t cross join
     (select rand() as r) r
where r.r >= cume_p - p and
      r.r < cume_p;

我没有在我的机器上安装mysql，所以这是未经测试的，但我认为这个一般的想法将工作。

SELECT Event
      FROM Your_Table
     WHERE CASE WHEN Event = 'A' THEN 
                CASE WHEN RAND() <= .25 THEN 1
                     END
                WHEN Event = 'B' THEN 
                CASE WHEN RAND() <= .1 THEN 1
                     END
                WHEN Event = 'C' THEN 
                CASE WHEN RAND() <= .15 THEN 1
                     END
                WHEN Event = 'D' THEN 
                CASE WHEN RAND() <= .5 THEN 1
                     END
            END = 1;

在应用程序编程语言（如java、python、c、php、javascript或您正在使用的其他语言）中，这应该很容易计算。您只需选择应用程序中的所有行，然后在易于编写的地方进行计算。
如果没有应用程序需要在数据库中执行，那么就不要这样做。为正确的工作使用正确的工具。数据库首先用于持久性，而不是计算。
另请参见xy问题。

如果您只想选择一个概率等于百分比的字段
我认为这样做很好：设置@mybound:=rand（）100；选择from event where chance\u percentage<@mybound order by chance\u percentage desc limit 1