In ArcGIS 10.1 my export of a joined table has more records than the original feature class from which it came. It took me a bit to figure out but it looks like if you have a duplicate in the field in which you are joining to, say a table for instance, the export feature class of the joined data will create a duplicate polygon in your feature class.
For instance, let�??s say you have a polygon feature class with just 1 record and you join this data to a table of 5 records that is a duplicate of each in the join field you are using. After the join you still have 1 record in your feature class table. However, I'm finding that when you export the joined data, I'm now left with a feature class containing 5 records where it made a duplicate polygon to match the number of records in the table.
I can't see why this would be intentional and is definitely not the output you get prior to ArcGIS 10.
It sounds like you are running into this NIM. This was rejected as by design, the reason being that outer joins are now executed in the database.
One was to get around this would be:
Export FC to shapefile > Export table to dbf > conduct join > export joined shapefile to new shapefile > import into file or personal GDB
I can think of many better ways of avoiding the expansion than creating a shapefile and dBase table!
1 Your could use the tool MakeQueryTable http://resources.arcgis.com/en/help/main/10.2/index.html#//00170000006r000000which allows a more sophisticated SQL query to manage the type of join. There are limitations with this tool, the most difficult is that all the tables in the query must be in the same geodatabase, so you might have to make a temporary copy.