On 2/22/06, Freeman, Donald <dofreeman@xxxxxxxxxxx> wrote: > > I'm guessing you wouldn't have check everything. You should be able to > check 10-15% of the data to make a valid statistical conclusion that the > tables are the same or not the same. As far as the character data you could > probably transform it to numerical data and then summarize it. Or, like > some others said, export the data and do file comparison on the export > files. > > That would depend on the reason for the check, and the amount of DML that may take place. If the check is for security, 10-15% probably wouldn't be sufficient. Then again, neither is comparing the number of rows. -- Jared Still Certifiable Oracle DBA and Part Time Perl Evangelist