declare @st datetime,
        @et datetime,
        @MaxT    datetime
set @st = '2012-08-15 00:00:00.000'
set @et = '2013-08-31 00:00:00.000' 
set @MaxT = @st
while(@maxt < @et)
begin
    set @MaxT = dateadd(mm,1,@maxt)
end
if (@MaxT > @et)
begin
    set @MaxT = dateadd(mm,-1,@maxt)
end

select DATEDIFF (mm, @st ,@MaxT )+ 
datediff(day,@MaxT,@et)*1.0/day(dateadd(mm,1,@MaxT)-day(@MaxT))

/*
---------------------------------------
12.516129032258

(1 行受影响)

*/

------解决方案--------------------

SQL code
declare @a datetime,
        @b datetime,
        @c datetime
set @a = '2012-08-15 00:00:00.000'
set @b = '2013-08-01 00:00:00.000'
set @c = '2013-08-31 00:00:00.000' 
 
--假设@b是恒大于@a的，大的多少未知
SELECT CASE WHEN DATEADD(mm,DATEDIFF(mm,@a,@b),@a)>=@b
THEN LTRIM(DATEDIFF(mm,@a,@b)-1)+'.'+LTRIM(DATEDIFF(dd,DATEADD(mm,DATEDIFF(mm,@a,@b)-1,@a),@b))
ELSE LTRIM(DATEDIFF(mm,@a,@b))+'.'+LTRIM(DATEDIFF(dd,DATEADD(mm,DATEDIFF(mm,@a,@b),@a),@b))
END 

SELECT CASE WHEN DATEADD(mm,DATEDIFF(mm,@a,@c),@a)>=@c
THEN LTRIM(DATEDIFF(mm,@a,@c)-1)+'.'+LTRIM(DATEDIFF(dd,DATEADD(mm,DATEDIFF(mm,@a,@c)-1,@a),@c))
ELSE LTRIM(DATEDIFF(mm,@a,@c))+'.'+LTRIM(DATEDIFF(dd,DATEADD(mm,DATEDIFF(mm,@a,@c),@a),@c))
END 

/*

-------------------------
11.17

(1 行受影响)


-------------------------
12.16

(1 行受影响)


*/

------解决方案--------------------

SQL code

declare @A datetime,@B datetime,@C datetime
set @A='2012-08-15 00:00:00.000'
set @B='2013-08-01 00:00:00.000' 
set @C='2013-08-31 00:00:00.000' 

select convert(decimal(18,1),datediff(mm,convert(varchar(7),@A,120)+'-01',convert(varchar(7),@B,120)+'-01'))
+convert(decimal(18,1),(day(@B)-day(@A))/30.0) B减A
,convert(decimal(18,1),datediff(mm,convert(varchar(7),@A,120)+'-01',convert(varchar(7),@C,120)+'-01'))
+convert(decimal(18,1),(day(@C)-day(@A))/30.0) C减A

/*
B减A                                     C减A
--------------------------------------- ---------------------------------------
11.5                                    12.5

(1 row(s) affected)

*/