ANSI SQL does not stipulate that nulls should be sorted at the low end of a collation. All that is said is that all nulls must be sorted the same way, either at the high end or low end. The actual choice is implementation defined, i.e. at the discretion of the vendor.<BR><BR>If you wish to change the behaviour you could add this quirk to your query, assuming that c1 is the column nulls that should be sorted at the end<BR><BR>select c1,c2,c3,case when c1 is null then 1 else 0 end as ed<BR> from t<BR> order by ed<BR><BR>If a DBMS does not support the case expression (which is standard) there is often a mapping function such as IIF, DECODE or VALUE et.c. that could be used.