Don't know how accurate you want your model, heres a simple but effective one...
On each bounce the ball loses energy so it dosen't bounce as high as it fell from i.e
if(bounce()){ yVel = -efficiency*yVel; }
where efficiency is some factor like 0.75. If you want the ground to have friction do
if(bounce()){xVel = friction*xVel; }
again friction is some factor like 0.9.
Hope that helps.

Thats a good question - I suggest you try both to see which looks better (you've probably aready done that though).
-zaph

wolfman8k is right except for when the object is at rest. friction is a reactive force, so it doesn't take effect until another force comes into play for it to act against. otherwise the ball would never come to rest either.
once the applied force is greater than the force of friction, the object moves and the total force becomes the force of acceleration minus the force of friction, which can be made constant for simplicity's sake and look very realistic.
but, once the object is in motion, if the applied force of acceleration matches the force of friction, the object moves at a constant rate. so, depending on how you want to make your game physics work, you may want to increase friction as velocity increases, as suggested by julian. this is how a 'maximum speed' is reached - at some point the force of friction matches the maximum force that can be applied for acceleration.
once the force of acceleration becomes less than the force of friction, the object slows down.
make sense? i hope so, i do ramble on so badly sometimes...
good luck!
jase

Yep, this has been discussed before, and I prevailed :-P
Just one thing to add: don't confuse friction with air resistance (sometimes also called friction or air friction). With air resistance, the opposing force is (generally) proportional to the square of the velocity. You'll have to be careful how you apply this, since the discreteness of computer maths could send the ball shooting off in the opposite direction!