>OK George it's different from ANSI SQL. I can understand it. I can't understand why if operand is from first table it returns records. I expected it to either return none or all in both cases.

Cetin,

I don't know Cetin. I just tried it on a parameterized remote view against SQL Server 7.0, using datetimes. The start datetime was later than the stop. No records came back. When I reversed them 722 came back. So I think it may apply to ANSI as well.

>There is a Turkish saying :
>'A dumb throwed a stone in the well and 40 wise couldn't take it of' :)
>I think this is kinda situation. MS team should think of it as a feature :) BTW this is under VFP5 and didn't test under VFP6 yet. Maybe it's in the fixed list.

Mine was done under VFP 6.0 so I don't think that it's viewed as a bug. I think my code snippet may be accurate. At least it explains what's happening.:-)