如果您不需要输出中的排名（并且它在您的评论和喜欢的答案中显示出来，而您并不需要），则可以简单地组合最接近Jan得分的测验得分：
查询（此处为SQL Fiddle）：
-- XXX this assumes `scores`.`username` is UNIQUE !SELECt * FROM ( -- Get those who scored worse (or tied) ( SELECt s.*FROM scores s CROSS JOIN (SELECt score FROM scores WHERe username = 'Jan') ref WHERe s.score <= ref.score AND username <> 'Jan' ORDER BY s.score DESC LIMIT 2) UNIOn -- Get our reference point record (SELECt s.* FROM scores s WHERe username = 'Jan') UNIOn -- Get those who scored better ( SELECt s.*FROM scores s CROSS JOIN (SELECt score FROM scores WHERe username = 'Jan') ref WHERe s.score > ref.score AND username <> 'Jan' ORDER BY s.score ASC LIMIT 2)) sliceORDER BY score ASC;
（请注意，由于示例数据集太小，我将结果限制为1月之前的两个记录和1月之后的两个记录。）
为了使LIMIT和UNIOn可以一起使用，需要在上面的组成查询上加上这些括号。然后，最外面的查询使我们可以对UNIOn的结果进行ORDER排序。
