When doing stuff commercially, transferring between applications usually uses some common exchange file format (omf etc) that works well between the packages in question. Even with these interchage formats you quickly find that each program's implementation of the file formats leads to a 'learning curve' when exchanging files between applications. Meaning even with universal interchange standards you must learn how to prepare what you're exporting, what options to set, what metadata to tag with, how to name clips, files and etc. So basically even with a direct 'built in' migration path it's a process of trial and error to get things to transfer reliably.
This doesn't directly apply to Cubase 5.1 > Live I don't think since I'm fairly sure Ableton Live doesn't support much for importing (beyond raw tracks you have figured), but the idea is still relevant here. Ie, you're going to have to pick a few projects and experiment to get the process down. You already have a basic idea of what you need, the midi structure & audio structure should all line up through the export>import process, and most plugin settings will have to be tested to see if they will work (via fxp/fxb). Now I can't comment directly on what you'll encounter with Cubase as I can't remember 5.1 much these days, but I can help with some hints on what might help for the audio & midi tracks:
For the MIDI bits, pick a common point at the start where there isn't *anything* and insert a single note & controller value on midi tracks to 'initalize' the starting point. That means you'll have to pad the beginning of your song with some meaningful interval (16 bars, 32bars, 8 bars or whatever u deem necessary) for ableton's sake. Quantize these bits of midi information it to exactly the start of that bar.
For extra credit any external gear can have sysex data recorded or cc values initialized after this point in its own track, assuming there is any (this was a common thing to do in the 'old days' during any tune's "preroll" or "lead-in" to get a device's settings dialed in).
For the audio you have several considerations as well, the first being your start point. The easiest thing to do here is to draw a waveform by hand or create it somewhere, I would suggest something like a saw wav, long enough to be visible from a decent zoom, ie, a single cycle from 0 to close to full volume (vertical line) that decends back down:
example done in Cubase SX:
http://www.kief.net/studio/exporting_wi ... points.jpg
Any serious level changes versus the raw audio may need to be applied to the track or you'll need the ability to do extensive comparing of your mixing automation once in Ableton to achieve the same balance. This also brings another point up, should existing eq & effects be applied or not? You have the option of with effects, without, and even better choosing some combination thereof. For instance you might mix several sections to completed 'stems' for easy remixing, then each track can be provided 'dry' and 'wet' with everything applied (and then should the 'dry' track be sans everything including volume or not?). It's entirely up to how robust you want your options to be later on...
After some thinking & experimenting with those options, you'll also have to figure out a meaningful naming convention for what you're rendering out so that things make sense on bringing the data back into Ableton. You might also want to experiment with 24bit and 32bit file formats to see what you're happiest with there (I think both applications support 32bit int & flt file formats, but not sure about Cubase 100%).