question

So, TableA on the publisher has n rows while the same TableA on the subscriber has >n rows and you want to move the extra rows from the subscriber up to the publisher? Are the n rows identical between the servers, or are there differences between shared rows? Replication, by itself, won't get your tables synchronized. When you set up replication, the publisher and subscriber have to be synchronized or you're stuck with your current problem. Of course, if you just import the rows into the publisher using SSIS it will immediately try to publish the rows back to the subscriber, so you'll have to turn off replication while you synchronize the tables. Once you have the data synchronized between the servers (SSIS is a prime candidate for this), you can set up replication again and the two tables will stay synchronized, just not with vanilla transactional replication. This is a classic case for merge replication and I would be interested to know why it is not an option. Without merge replication, you are left with *Transactional Replication with Updatable Subscriptions* which has been deprecated, but will get you going until you can write a replacement - [ http://technet.microsoft.com/en-us/library/ms151718.aspx][1] - or Peer-to-peer replication (an enterprise edition feature) between your two servers - [ http://technet.microsoft.com/en-us/library/ms151196.aspx][2] [1]: http://technet.microsoft.com/en-us/library/ms151718.aspx [2]: http://technet.microsoft.com/en-us/library/ms151196.aspx