I've just picked up a new client who saves everything in 16-bit .tif files. What format should I chose to work in? These puppies are HUGE! It is taking me 24 hours on DSL to download all of his photos from ftp.

grafx, there seem to be two questions here:

1) jpg / tif ?
2) 16 bit / 8 bit ?

Question 1 is easy, jpg never.

Question 2 is trickier.
Today*, I would have to reply always work in 8 bits. The processing capacity, tools and memory normally available just aren't geared to 16 bits. If you have CS2, you can sortta work in 16 bits but often the tool you need won't be available.
There has been some discussion, with some claiming that 16 bits is just a complete waste of time (and space), because we humans (and that who it's all for, after all) can't see any difference between the two types of images - even after heavy processing.
The only place where you should always use higher bit depths is in image capturing. A camera RAW file uses 12 bits (as I remember), a scanner can give 16 bits, and a HDR image can have 24 or 32 bits - each gives a significantly higher range of luminosities than you can print ,or your monitor can show (our your eye can register).
This extra input information enables you to choose a range in which you want to concentrate. For all of these, you will do some initial simple processing in the native format and then reduce to 8 bits for processing the "visible" image.

Summing up: Do some levels adjustments in 16 bits, then convert to 8.

*In a few years time, we'll be working on quantum computers, using flash memories, and this worrying about more than 8 bits will be quite laughable.

Rô