The main value of a hand-held meter is that it measure the amount of light falling on a scene (or object), independent of the reflectance or tone of the object. So let us assume we have a large dark object (say a dog) surrounded by bright objects (say, sand on a beach). You could take an incident light meter reading of the scene and get something like 1/100 at f/16 at ISO 100 (this happens to be the sunny-16 exposure rule). If you took a spot-meter reading off the dog, chances are the meter would require an exposure of maybe 1/10 sec because it is trying to get the black of the dog to look medium grey. If you took a spot-meter reading of the white sand, the chances are you've have a reading requiring 1/500 sec at f/16, because the meter again will be trying to adjust the exposure so that the result is medium grey. Of course, neither of the two exposures will be correct, because the dog "should" look black, and the sand "should" look white. If you're using only the reflected light meter (which all camera meters are), you would need to adjust the exposure compensation to "tell" the camera that the reading off the dog should be 3 stops darker than what it is trying to do. If you were using the spotmeter reading off the sand, you want to keep the sand white (not gray), and you will need to dial in an exposure compensation of about +2 to have the sand show up as white. Using an incident light meter, you don't have to fool around with exposure compensation, because it doesn't care about the actual reflectance of the objects in the scene.
However, these are just tools to allow you do get a "proper" image. Don't forget as well, that the camera has a limited dynamic range. My own camera (Canon T1i) has an apparent range of -4 and +3 (it takes 4 stops to change a medium grey object to solid black, and 3 stops to change it to solid white), and whatever detail I want to see in the image must fit into that range (that is, without using additional tools like HDR, tone-mapping, blending, etc.). So in my case I take a spot-meter reading off a highlight where I want detail, and assuming it is a white object, I'll increase the exposure by 2 stops to give me the "correct" exposure. If the important stuff is in the shadows, then I take a reading off the shadow detail and reduce the exposure by 3 stops. If the dynamic range of the scene exceeds 7 stops, then I have to use blending or HDR techniques to squish the dynamic range into one the image can show.
There are times when there are different light sources and they interact. You could have something in direct sunlight, with a portion reflected off a nearby object, and a portion partly in shade and lit only by the overhead sky. So there are at least three light sources that I need to take into account when figuring out the exposure. The simplest way is just to use the incident light meter to read the light at each relevant point, and then make a decision on how to bias the exposure setting. If I can't get there with an incident light meter, then I would take spot-meter readings, and adjust for the tonality/reflectance of the objects. Once I have an appropriate exposure, I'll take a test shot and check the histogram to see where the clipping/blocking appears and how the overall scene is distributed. From there, I'd adjust. Once I've dialed in my settings, I generally don't look at the histogram again unless the light changes.