I have been in the rather unique position of recently owning both
a Kenwood TS790 and a Yaesu FT736.  I thought a couple of
comparative observations might be useful to those considering
both of these radios.  The following observations are based on
casual operating, not on scientific measurement.

I owned the 790 first having bought it this year at Dayton.  It
is an outstanding radio with an apparantly excellent receive and
more bells and whistles that you can shake a stick at.  The
reasons that I bought it initially rather than the 736 were that
it had somewhat higher power output, it displayed both operating
frequencies at the same time, and it had the ability to listen on
two bands at once.  If I were not a satellite operator, I would
probably still own the Kenwood; I thought it was a marvelous
radio.

However, my interest in the radio was limited almost exclusively
to satellite operation and I quickly found out that for this
purpose the radio was, from my point of view, unacceptable.  Let
me explain:

I operated that 790 at a friends house before going to Dayton (he
had just recently acquired it and had not figured it all out
himself).  I had problems getting the VFO's to track properly for
satellite operation.  The way the 790 works is to store in memory
the sum of frequencies.  Then with the push of the sat button you
can bring your transmit vfo up to wherever you happen to be
receiving at any given time.  That is, it subtracts you recieve
frequency from the sum of frequencies in memory and then puts
your xmit vfo on the difference between these.  

There is only one problem with this... as the doppler shifts the
sum of frequencies changes.  When I first called Kenwood to ask
about how to correct for this they said it was not necessary
since the doppler affected both up and downlink.  I pointed out
that the doppler was more severe on 440 and they said yes, but
not enough to worry about.  I concluded that the person I spoke
with had not spent much time on Oscar 13.

I asked the Kenwood technical person about this at Dayton and he
agreed at the point that it was necessary to compensate for the 
changing doppler, but insisted the radio could do it.  On the
strength of this I bought the 790.  I used it for about 3 days
at home and spent about an hour on the line with Kenwood. The
upshot is that there are only two ways to track a changing
doppler.  1. Add up the xmit and recv freqs with a pocket
calculator and re-enter them into memory (I might as well have
stayded with dual multi-mode radios!).  2. Kenwood insisted that
if you tuned to a vacant sum of frequencies memory and then held
in the SAT button while you tuned it would track whatever sum of
frequencies happened to be in the vfo's at the time.  I
personally couldn't get this to work, but even if it did work it
meant that you had to use 2 hands to tune the radio.

As a result, I concluded that the 790 was unacceptable for
routine phase 3 satellite communications.  That's when I bought
the 736.

The 736 also has an excellent receiver (both of these receivers
are better than anything I've ever heard before).  It doesn't
have all the bells a whistles of the 790 but it does have dual
pairs of vfo's for satellite work (I leave one on mode J and one
on mode B) and 10 dedicated satellite memories.  In short, if I
were looking to buy a radio for other than satellite uses I would
probably buy the Kenwood.  However, the 736 seems to be designed
with the satellite operator in mind and is much superior in that
use  73, John WA0PTV @WA0PTV