Optoisolators are very simple. If you"re using them as switches, you can basically model them as LEDs and drive them with a series current limiting resistor just like you would any other LED. Set the current from the datasheet"s LED current vs. collector current graph so that you have the necessary output sinking capability that you need.
On the output side, they are usually NPN transistors, and you"ll normally at least have access to the collector and emitter. So either configure them in common collector or common emitter depending on your needs. Available source/sink current will be provided in the datasheet as a curve of diode current vs. collector current. With a typical 4N35, it"s kind-of linear and close to 1:1, so you shouldn"t need an extra driver vs. driving directly from the micro pins. If you need push-pull output (ie. need to both source and sink current), you'll need to buffer the output of the optoisolator somehow, using 74 series logic is probably the easiest way.
If we"re only talking about slow-speed requirements (say, < 10KHz), you can basically neglect everything else.
That said, unless you want separate ground domains for some reason, there"s not really much reason to use them.
Edit: Okay I see you want to use these in a vehicle, that's a good application for them. The easiest thing to do is to use a low LED current (say 1mA), such that relatively high voltages don't come anywhere close to exceeding the specification. Since you're driving it into a microcontroller you can use large pull-up resistors and will need very little output current capability.