Why doesn't MRI show soft-tissue injuries well?  A good analogy is a car engine.  Let's say your car didn't start one morning.  You break out your trusty Polaroid and take a few snapshots of the engine.  You then hitch a ride to your mechanic and show him or her the pictures.  How likely is it that the mechanic will be able to diagnose your car based solely on those pictures?  Not very.

MRI shows structure, just like the Polaroid's.  So if there's something drastically wrong (with your car-like oil leaking or a tree that's fallen on the hood, or with your spine-like a herniated disc or a bone spur pressing on a nerve), it's a good test.  However, there are 5,000 other things that can prevent your car from starting that would never show up on the Polaroid's.  Since you spine is several million times more complex than your car, picture type tests that look only at structure aren't usually very helpful in telling us what's wrong.

Going back to the mechanic analogy, how would your mechanic diagnose the problem?  He or she would ask you questions about how it started, open the hood and poke around, look for tell tale signs of problems that are common, ask you about anything unusual you've noticed, etc...  In the same way, your doctor should use the MRI as a piece of a big puzzle, not a way to diagnose or exclude an injury.