I'm surprised that anyone working with SQL Server needs lessons in removing duplicates, and puzzled at the complexity of this cursor-based solution originally published. A better solution has already been pointed out in the forum. As to the opinion that databases should never contain duplicate entries, yes, but somehow they sometimes creep in. Like the Kernighan and Richie exception error message, 'this is impossible', which one should never see and often does. Commercial databases are often of frightening complexity and one should be always on guard for duplicates. Yes, constraints are one way of doing it, but there are times when the overhead is too great.