Values stored in partitioned tables are unloaded into files with one partition per file using a database unload utility. This utility forms a punch file that includes instructions to perform a load of data from the unloaded files into the partitioned table. The punch file is read to determine the names of the unloaded files. These names are altered to form new files. Values in the unloaded files are modified and the modified values are written to the new files. A new punch file is created that is more efficient than the initial punch file and that allows data in the partitioned table to be replaced on a partition basis with the data in the new files. The new punch file is then executed to load the modified values in the new files into the partitioned table.