The bounding box and the collision meshes are 2 different things, and each serves a different purpose. What Duion says is true, in a way, but only when dealing with the actual player character that the client controls. As I recommended in your other posting @ saindd
you should setup the player's bounding box dimensions in the player's datablock file. There are fields there for the bounding box and crouching bounding box, etc.
The player is an exception to the rule, in that the bounding box dimensions setup in the player's datablock will also be the player's collision box. Other objects actually use bounding boxes and collision meshes as they 'normally' should be - The bounding box determines visible bounds and the collision meshes determine actual collision. You can see how this works pretty easily:
• Create a large cube taller than the player in size.
• Create a bounds box for the cube, say half the size of the cube so that the bounds box is too small and fits inside the cube.
Toss the cube into a scene and look at it. If you turn the camera to the side, the cube will disappear when the bounds box goes out of view. This is totally incorrect because you'll notice when the cube is not in view and you turn to look at it again it won't render until the bounds box is on screen again. So you end up with a cube that should be rendered but isn't until you're looking closer to its center.
Collision meshes are pretty self explanatory, quite simply they are what you actually collide with. I just wanted to point out the functional differences between the two.