alter table MyTable add sno int identity(1,1)
    delete from MyTable where sno in
    (
    select sno from (
    select *,
    RANK() OVER ( PARTITION BY RowID,Col3 ORDER BY sno DESC )rank
    From MyTable
    )T
    where rank>1
    )

    alter table MyTable 
    drop  column sno

这个查询为我展示了非常好的性能:

DELETE tbl
FROM
    MyTable tbl
WHERE
    EXISTS (
        SELECT
            *
        FROM
            MyTable tbl2
        WHERE
            tbl2.SameValue = tbl.SameValue
        AND tbl.IdUniqueValue < tbl2.IdUniqueValue
    )

它在30秒多一点的时间内从2M的表中删除了1M行(50%重复)

哦,当然。使用临时表。如果你想要一个“工作”的单一的、性能不太好的语句，你可以使用:

DELETE FROM MyTable WHERE NOT RowID IN
    (SELECT 
        (SELECT TOP 1 RowID FROM MyTable mt2 
        WHERE mt2.Col1 = mt.Col1 
        AND mt2.Col2 = mt.Col2 
        AND mt2.Col3 = mt.Col3) 
    FROM MyTable mt)

基本上，对于表中的每一行，子选择将查找与所考虑行的完全相同的所有行的顶部RowID。因此，您最终会得到一个表示“原始”非重复行的RowIDs列表。

DELETE 
FROM MyTable
WHERE NOT EXISTS (
              SELECT min(RowID)
              FROM Mytable
              WHERE (SELECT RowID 
                     FROM Mytable
                     GROUP BY Col1, Col2, Col3
                     ))
               );

我知道这个问题已经回答了，但我已经创建了非常有用的sp，它将为表副本创建一个动态删除语句:

    CREATE PROCEDURE sp_DeleteDuplicate @tableName varchar(100), @DebugMode int =1
AS 
BEGIN
SET NOCOUNT ON;

IF(OBJECT_ID('tempdb..#tableMatrix') is not null) DROP TABLE #tableMatrix;

SELECT ROW_NUMBER() OVER(ORDER BY name) as rn,name into #tableMatrix FROM sys.columns where [object_id] = object_id(@tableName) ORDER BY name

DECLARE @MaxRow int = (SELECT MAX(rn) from #tableMatrix)
IF(@MaxRow is null)
    RAISERROR  ('I wasn''t able to find any columns for this table!',16,1)
ELSE 
    BEGIN
DECLARE @i int =1 
DECLARE @Columns Varchar(max) ='';

WHILE (@i <= @MaxRow)
BEGIN 
    SET @Columns=@Columns+(SELECT '['+name+'],' from #tableMatrix where rn = @i)

    SET @i = @i+1;
END

---DELETE LAST comma
SET @Columns = LEFT(@Columns,LEN(@Columns)-1)

DECLARE @Sql nvarchar(max) = '
WITH cteRowsToDelte
     AS (
SELECT ROW_NUMBER() OVER (PARTITION BY '+@Columns+' ORDER BY ( SELECT 0)) as rowNumber,* FROM '+@tableName
+')

DELETE FROM cteRowsToDelte
WHERE  rowNumber > 1;
'
SET NOCOUNT OFF;
    IF(@DebugMode = 1)
       SELECT @Sql
    ELSE
       EXEC sp_executesql @Sql
    END
END

如果你创建这样的表格

IF(OBJECT_ID('MyLitleTable') is not null)
    DROP TABLE MyLitleTable 


CREATE TABLE MyLitleTable
(
    A Varchar(10),
    B money,
    C int
)
---------------------------------------------------------

    INSERT INTO MyLitleTable VALUES
    ('ABC',100,1),
    ('ABC',100,1), -- only this row should be deleted
    ('ABC',101,1),
    ('ABC',100,2),
    ('ABCD',100,1)

    -----------------------------------------------------------

     exec sp_DeleteDuplicate 'MyLitleTable',0

它将从表中删除所有重复项。如果运行它时不带第二个参数，它将返回一条SQL语句来运行。

如果您需要排除任何列，只需在调试模式下运行它，获取代码并按照您的喜好修改它。

另一种可能的方法是

; 

--Ensure that any immediately preceding statement is terminated with a semicolon above
WITH cte
     AS (SELECT ROW_NUMBER() OVER (PARTITION BY Col1, Col2, Col3 
                                       ORDER BY ( SELECT 0)) RN
         FROM   #MyTable)
DELETE FROM cte
WHERE  RN > 1;

我在上面使用ORDER BY (SELECT 0)，因为在出现平局的情况下，保留哪一行是任意的。

例如，要以RowID顺序保存最新的一个，您可以使用order BY RowID DESC

执行计划

它的执行计划通常比接受的答案更简单和更有效，因为它不需要自连接。

然而，情况并非总是如此。GROUP BY解决方案可能会优先于选择散列聚合而不是流聚合的情况。

ROW_NUMBER解决方案总是给出几乎相同的计划，而GROUP BY策略则更加灵活。

可能有利于哈希聚合方法的因素是

分区列上没有有用的索引 相对较少的组，每组的重复数相对较多

在第二种情况的极端版本中(如果每个组中有很多重复的组)，还可以考虑简单地插入要保留到新表中的行，然后截断原始的行并将它们复制回来，以最大限度地减少日志记录，而不是删除非常高比例的行。