刚从Oracle转到SQL Server确实会遇到语法适配的问题，Oracle的(+)外连接语法在SQL Server里不被支持，不过我们可以用标准的LEFT JOIN（和LEFT OUTER JOIN等价，SQL Server中两者通用）来实现完全相同的逻辑。

先拆解你的原Oracle查询：核心是主表DTE24左连接三个聚合子查询totaal、totaal1、totaal2，连接条件是CONTROLE和WAARDEBEG字段匹配。下面是改写后的SQL Server兼容版本：

SELECT DISTINCT
    DTE24.CONDSRT,
    DTE24.CONTROLE,
    DTE24.WAARDEBEG,
    totaal.totaalx,
    totaal1.totaalof1,
    totaal2.totaalof2
FROM [RT_IPOIKAZ_ODS].[WinIKAZ].[DTE24]
LEFT JOIN (
    SELECT 
        CONTROLE, 
        WAARDEBEG, 
        COUNT(WAARDEBEG) AS totaalx
    FROM [RT_IPOIKAZ_ODS].[WinIKAZ].[DTE24]
    WHERE CONTROLE = 12885 
      AND DATAFVOER = 0 
      AND CONDSRT = 'PRESCODE' 
      AND OFNIVO IN (1,2)
    GROUP BY CONTROLE, WAARDEBEG
) AS totaal 
    ON DTE24.CONTROLE = totaal.CONTROLE 
    AND DTE24.WAARDEBEG = totaal.WAARDEBEG
LEFT JOIN (
    SELECT 
        CONTROLE, 
        WAARDEBEG, 
        COUNT(WAARDEBEG) AS totaalof1
    FROM [RT_IPOIKAZ_ODS].[WinIKAZ].[DTE24]
    WHERE CONTROLE = 12885 
      AND DATAFVOER = 0 
      AND CONDSRT = 'PRESCODE' 
      AND OFNIVO = 1
    GROUP BY CONTROLE, WAARDEBEG
) AS totaal1 
    ON DTE24.CONTROLE = totaal1.CONTROLE 
    AND DTE24.WAARDEBEG = totaal1.WAARDEBEG
LEFT JOIN (
    SELECT 
        CONTROLE, 
        WAARDEBEG, 
        COUNT(WAARDEBEG) AS totaalof2
    FROM [RT_IPOIKAZ_ODS].[WinIKAZ].[DTE24]
    WHERE CONTROLE = 12885 
      AND DATAFVOER = 0 
      AND CONDSRT = 'PRESCODE' 
      AND OFNIVO = 2
    GROUP BY CONTROLE, WAARDEBEG
) AS totaal2 
    ON DTE24.CONTROLE = totaal2.CONTROLE 
    AND DTE24.WAARDEBEG = totaal2.WAARDEBEG
WHERE DTE24.CONTROLE = 12885 
  AND DTE24.DATAFVOER = 0 
  AND DTE24.CONDSRT = 'PRESCODE'
ORDER BY DTE24.WAARDEBEG;

关键改写点说明： #

• 替换(+)为LEFT JOIN：Oracle中DTE24.WAARDEBEG = totaal.WAARDEBEG(+)等价于SQL Server的DTE24 LEFT JOIN totaal ON DTE24.WAARDEBEG = totaal.WAARDEBEG，这样主表DTE24的所有符合条件的行都会被保留，即使子查询中没有匹配的记录，对应的聚合字段会显示为NULL。
• 明确连接条件：每个LEFT JOIN都单独指定连接的字段（CONTROLE和WAARDEBEG），这比Oracle的隐式连接写法更清晰，也符合SQL标准。
• 补全主表过滤条件：我给主表DTE24加上了和子查询一致的过滤条件，这能确保你拿到的数据集和原Oracle查询逻辑完全一致——原Oracle查询里虽然没写主表的这些过滤，但子查询的条件加上连接关系，其实隐含了主表的这些过滤，明确写出来更直观。
• ORDER BY优化：把原查询的ORDER BY 3改成了ORDER BY DTE24.WAARDEBEG，用列名排序比用位置更易读，也避免后续修改列顺序时出错。

如果运行后有聚合结果不符合预期的情况，可以检查子查询的过滤条件是否和原Oracle一致，或者主表的过滤是否正确。

内容的提问来源于stack exchange，提问作者NiekP