2019 UPDATE: In the 10 years since I wrote this answer, more solutions have been uncovered that may yield better results. Also, SQL Server releases since then (especially SQL 2012) have introduced new T-SQL features that can be used to calculate medians. SQL Server releases have also improved its query optimizer which may affect perf of various median solutions. Net-net, my original 2009 post is still OK but there may be better solutions on for modern SQL Server apps. Take a look at this article from 2012 which is a great resource: https://sqlperformance.com/2012/08/t-sql-queries/median

本文发现，以下模式比所有其他选择都要快得多，至少在他们测试的简单模式上是这样。该解决方案比测试的最慢解决方案(PERCENTILE_CONT)快373x (!!)注意，这个技巧需要两个独立的查询，这可能不是在所有情况下都可行。它还需要SQL 2012或更高版本。

DECLARE @c BIGINT = (SELECT COUNT(*) FROM dbo.EvenRows);

SELECT AVG(1.0 * val)
FROM (
    SELECT val FROM dbo.EvenRows
     ORDER BY val
     OFFSET (@c - 1) / 2 ROWS
     FETCH NEXT 1 + (1 - @c % 2) ROWS ONLY
) AS x;

当然，仅仅因为2012年对一个模式的一次测试产生了很好的结果，您的实际情况可能会有所不同，特别是如果您使用的是SQL Server 2014或更高版本。如果性能对中值计算很重要，我强烈建议尝试并测试那篇文章中推荐的几个选项，以确保您找到了最适合您的模式的选项。

我还会特别小心地使用(SQL Server 2012新增的)函数PERCENTILE_CONT，这是这个问题的其他答案之一中推荐的，因为上面链接的文章发现这个内置函数比最快的解决方案慢373x。在过去的7年里，这种差异可能已经得到了改善，但就我个人而言，在验证它与其他解决方案的性能之前，我不会在大型表上使用这个函数。

2009年的原始帖子如下:

有很多方法可以做到这一点，它们的性能差别很大。下面是一个优化得特别好的解决方案，包括median、ROW_NUMBERs和性能。当涉及到执行期间生成的实际I/ o时，这是一个特别优的解决方案——它看起来比其他解决方案成本更高，但实际上要快得多。

该页还包含对其他解决方案和性能测试细节的讨论。请注意，如果有多行具有相同的中位数列值，则使用唯一列作为消歧器。

就像所有的数据库性能场景一样，总是尝试在真实的硬件上用真实的数据测试解决方案——你永远不知道什么时候对SQL Server优化器的更改或环境中的某个特性会使正常快速的解决方案变慢。

SELECT
   CustomerId,
   AVG(TotalDue)
FROM
(
   SELECT
      CustomerId,
      TotalDue,
      -- SalesOrderId in the ORDER BY is a disambiguator to break ties
      ROW_NUMBER() OVER (
         PARTITION BY CustomerId
         ORDER BY TotalDue ASC, SalesOrderId ASC) AS RowAsc,
      ROW_NUMBER() OVER (
         PARTITION BY CustomerId
         ORDER BY TotalDue DESC, SalesOrderId DESC) AS RowDesc
   FROM Sales.SalesOrderHeader SOH
) x
WHERE
   RowAsc IN (RowDesc, RowDesc - 1, RowDesc + 1)
GROUP BY CustomerId
ORDER BY CustomerId;

--Create Temp Table to Store Results in
DECLARE @results AS TABLE 
(
    [Month] datetime not null
 ,[Median] int not null
);

--This variable will determine the date
DECLARE @IntDate as int 
set @IntDate = -13


WHILE (@IntDate < 0) 
BEGIN

--Create Temp Table
DECLARE @table AS TABLE 
(
    [Rank] int not null
 ,[Days Open] int not null
);

--Insert records into Temp Table
insert into @table 

SELECT 
    rank() OVER (ORDER BY DATEADD(mm, DATEDIFF(mm, 0, DATEADD(ss, SVR.close_date, '1970')), 0), DATEDIFF(day,DATEADD(ss, SVR.open_date, '1970'),DATEADD(ss, SVR.close_date, '1970')),[SVR].[ref_num]) as [Rank]
 ,DATEDIFF(day,DATEADD(ss, SVR.open_date, '1970'),DATEADD(ss, SVR.close_date, '1970')) as [Days Open]
FROM
 mdbrpt.dbo.View_Request SVR
 LEFT OUTER JOIN dbo.dtv_apps_systems vapp 
 on SVR.category = vapp.persid
 LEFT OUTER JOIN dbo.prob_ctg pctg 
 on SVR.category = pctg.persid
 Left Outer Join [mdbrpt].[dbo].[rootcause] as [Root Cause] 
 on [SVR].[rootcause]=[Root Cause].[id]
 Left Outer Join [mdbrpt].[dbo].[cr_stat] as [Status]
 on [SVR].[status]=[Status].[code]
 LEFT OUTER JOIN [mdbrpt].[dbo].[net_res] as [net] 
 on [net].[id]=SVR.[affected_rc]
WHERE
 SVR.Type IN ('P') 
 AND
 SVR.close_date IS NOT NULL 
 AND
 [Status].[SYM] = 'Closed'
 AND
 SVR.parent is null
 AND
 [Root Cause].[sym] in ( 'RC - Application','RC - Hardware', 'RC - Operational', 'RC - Unknown')
 AND
 (
  [vapp].[appl_name] in ('3PI','Billing Rpts/Files','Collabrent','Reports','STMS','STMS 2','Telco','Comergent','OOM','C3-BAU','C3-DD','DIRECTV','DIRECTV Sales','DIRECTV Self Care','Dealer Website','EI Servlet','Enterprise Integration','ET','ICAN','ODS','SB-SCM','SeeBeyond','Digital Dashboard','IVR','OMS','Order Services','Retail Services','OSCAR','SAP','CTI','RIO','RIO Call Center','RIO Field Services','FSS-RIO3','TAOS','TCS')
 OR
  pctg.sym in ('Systems.Release Health Dashboard.Problem','DTV QA Test.Enterprise Release.Deferred Defect Log')
 AND  
  [Net].[nr_desc] in ('3PI','Billing Rpts/Files','Collabrent','Reports','STMS','STMS 2','Telco','Comergent','OOM','C3-BAU','C3-DD','DIRECTV','DIRECTV Sales','DIRECTV Self Care','Dealer Website','EI Servlet','Enterprise Integration','ET','ICAN','ODS','SB-SCM','SeeBeyond','Digital Dashboard','IVR','OMS','Order Services','Retail Services','OSCAR','SAP','CTI','RIO','RIO Call Center','RIO Field Services','FSS-RIO3','TAOS','TCS')
 )
 AND
 DATEADD(mm, DATEDIFF(mm, 0, DATEADD(ss, SVR.close_date, '1970')), 0) = DATEADD(mm, DATEDIFF(mm,0,DATEADD(mm,@IntDate,getdate())), 0)
ORDER BY [Days Open]



DECLARE @Count AS INT
SELECT @Count = COUNT(*) FROM @table;

WITH MyResults(RowNo, [Days Open]) AS
(
    SELECT RowNo, [Days Open] FROM
        (SELECT ROW_NUMBER() OVER (ORDER BY [Days Open]) AS RowNo, [Days Open] FROM @table) AS Foo
)


insert into @results
SELECT 
 DATEADD(mm, DATEDIFF(mm,0,DATEADD(mm,@IntDate,getdate())), 0) as [Month]
 ,AVG([Days Open])as [Median] FROM MyResults WHERE RowNo = (@Count+1)/2 OR RowNo = ((@Count+1)%2) * ((@Count+2)/2) 


set @IntDate = @IntDate+1
DELETE FROM @table
END

select *
from @results
order by [Month]

虽然Justin grant的解决方案看起来很可靠，但我发现当您在给定的分区键中有许多重复值时，ASC重复值的行号最终会不按顺序排列，因此它们不能正确对齐。

以下是我的研究结果的一个片段:

KEY VALUE ROWA ROWD  

13  2     22   182
13  1     6    183
13  1     7    184
13  1     8    185
13  1     9    186
13  1     10   187
13  1     11   188
13  1     12   189
13  0     1    190
13  0     2    191
13  0     3    192
13  0     4    193
13  0     5    194

我使用Justin的代码作为这个解决方案的基础。尽管考虑到使用多个派生表效率不高，但它确实解决了我遇到的行排序问题。任何改进都会受到欢迎，因为我在T-SQL方面不是那么有经验。

SELECT PKEY, cast(AVG(VALUE)as decimal(5,2)) as MEDIANVALUE
FROM
(
  SELECT PKEY,VALUE,ROWA,ROWD,
  'FLAG' = (CASE WHEN ROWA IN (ROWD,ROWD-1,ROWD+1) THEN 1 ELSE 0 END)
  FROM
  (
    SELECT
    PKEY,
    cast(VALUE as decimal(5,2)) as VALUE,
    ROWA,
    ROW_NUMBER() OVER (PARTITION BY PKEY ORDER BY ROWA DESC) as ROWD 

    FROM
    (
      SELECT
      PKEY, 
      VALUE,
      ROW_NUMBER() OVER (PARTITION BY PKEY ORDER BY VALUE ASC,PKEY ASC ) as ROWA 
      FROM [MTEST]
    )T1
  )T2
)T3
WHERE FLAG = '1'
GROUP BY PKEY
ORDER BY PKEY

这段代码有点长，但很容易理解

medii是有列val的表，它有数据集， Smedi是一个cte，它将列idx作为行号，val作为medi表中的'val'，该表是升序排序的。 这是基本的数学，如果行号是奇数，那么它的中值来自smedi。 当它是偶数时，它是中间两个值的平均值。

with smedi(idx,vals) as(
                select ROW_NUMBER() over(order by val),val from medi
                )
select (case
            when (select count(*) from medi)%2!=0 then (select vals from smedi where (((select count(*) from medi)/2))=idx)
            else (select avg(vals) from smedi where idx in ((select count(*)/2 from medi),(select (count(*)/2)+1 from medi)))
            end)

以下解决方案在这些假设下有效:

无重复值 没有取消

代码:

IF OBJECT_ID('dbo.R', 'U') IS NOT NULL
  DROP TABLE dbo.R

CREATE TABLE R (
    A FLOAT NOT NULL);

INSERT INTO R VALUES (1);
INSERT INTO R VALUES (2);
INSERT INTO R VALUES (3);
INSERT INTO R VALUES (4);
INSERT INTO R VALUES (5);
INSERT INTO R VALUES (6);

-- Returns Median(R)
select SUM(A) / CAST(COUNT(A) AS FLOAT)
from R R1 
where ((select count(A) from R R2 where R1.A > R2.A) = 
      (select count(A) from R R2 where R1.A < R2.A)) OR
      ((select count(A) from R R2 where R1.A > R2.A) + 1 = 
      (select count(A) from R R2 where R1.A < R2.A)) OR
      ((select count(A) from R R2 where R1.A > R2.A) = 
      (select count(A) from R R2 where R1.A < R2.A) + 1) ; 

在Jeff Atwood的答案的基础上，它是用GROUP BY和一个相关的子查询来获得每个组的中位数。

SELECT TestID, 
(
 (SELECT MAX(Score) FROM
   (SELECT TOP 50 PERCENT Score FROM Posts WHERE TestID = Posts_parent.TestID ORDER BY Score) AS BottomHalf)
 +
 (SELECT MIN(Score) FROM
   (SELECT TOP 50 PERCENT Score FROM Posts WHERE TestID = Posts_parent.TestID ORDER BY Score DESC) AS TopHalf)
) / 2 AS MedianScore,
AVG(Score) AS AvgScore, MIN(Score) AS MinScore, MAX(Score) AS MaxScore
FROM Posts_parent
GROUP BY Posts_parent.TestID