Similarly for an UPDATE statement. Oracle must read the entire block into memory and then modify the row's contents. If the row increases in length due to the UPDATE statement and the row has to be migrated or chained, the statement will take longer to process. But this is true no matter how many columns are modified. Finally, it will be quicker to update all columns at once than to break the statement into multiple UPDATE statements.
The INSERT statement can take longer to complete the more data you insert. This has to do with the entire row length rather than the number of columns involved. But you shouldn't see any noticeable difference involved. If your INSERT statement is taking a long time, there are probably other factors at work.
This was first published in May 2008