Quality of timing simulation

Hello newsgroup users,

I have made a timing simulation for my SRAM-Controller and an external
asynchronous SRAM.

I have used the FPGA Cyclone timing output file for my SRAM-Controller
and the extra timing file for the SRAM model.
Both components have been instantiated in a testbench.


How reliable are such simulations? 
Where do come problems along with that kind of simulations ?  
Is it recommendable to simulate that way when connecting some
synchronous design with some asynchronous component?

I would appreciate your help or your field report about your timing
simulation.

Thanks a lot.

Rgds