You can start your program from any address right after then end of the highest addressed interrupt vector that is used by your application. However most seasoned 8051/8052 programmers avoid this practice and start their code after the end of the highest addressed interrupt vector that is supported by the particular processor they are using. In this way they can place some kind of valid code at the unused interrupt vectors. This will cause something sane to happen if an interrupt that is not expected does indeed happen. These unused interrupt vectors can be programmed in various ways. Some ideas are:
1) Program to simply return with the "return from interrupt" instruction.
2) Program it transfer to a generic trap handler that can report the invalid interrupt.
3) Program unused interrupt to invoke a soft reset by pointing to address zero.

Mike Karas