I don't use Sonar but the theory is pretty much the same with CUbase SX3 and Reaper I assume. Make sure you have 'snap' on so that if you drop in to repair a phrase or simply want to record each riff or section seperately everything will match up. Without snap things go out of time very easily and it is nearly impossible to edit anything properly.
If you are recording something very difficult then you are better off breaking it down to its component parts and recording in sections. Don't just play a riff once and copy it out to loop sections - that never sounds good. It's the little variations in timing, harmonics and transition notes that keep things interesting and organic. Get the part you want perfected then save it to track 1, then move on to the next part on the next track. Do as many takes as you have to to get the part right. When you have two perfect parts on seprate track then edit the beginning and end of each wav so the start and stop at exactly the same position. Then move them onto your master track (eg Rythym Guitar 1), highlight them and hit 'cross fade'. This will add a bit to the beginning and end of each part so they overlap and fade in and fade out of each other - make sure the cross fade is very short so that the transition sounds natural. That's how you do it. It doesnt really work that well for lead becuase everything is so precise - for lead and harmony parts just make sure that you only cross fade at the end of defined sections. Chopping and cutting throughout a set of arpeggios or in the middle of a shred run is almost impossible to do.
When you get a section down for 'Rhythm 1 Left', go straight about recording onto a fresh track 'Rhythm 1 Right'. Hard pan each track to give a beefy sound. The reason this is so important is that from the very beginning you will get a much better idea of what works and what doesnt, plus you can contruct the riffs on the fly to make them more interesting - add in 5ths or 3rds etc for rhythm harmonies just like Chuck used to do, whatever works. Using the same princiapl you can multitrack at various degrees of pan using reverb and other effects as plug-ins. Try to keep all of your effects for post recording as VST plug-ins, there is nothng worse than playing a part that sounds cool with a flanger or delay on that takes 150 attempts to get right only to discover that it doesnt work in the final mix.
Re volume differences etc... are you micing an amp or cab or re you going direct? It's much easier and more practical to go direct if you are recording over a number of days as an amp will sound different everytime (unless you have it set up in an isloation box but even then air temperature will affect the tone). If you are using a mic'ed amp even a 1cm change in the position of the mic will alter the tone. Also, I often notice that my picking hand will inadvertently hit the volume knob from time to time and the volume will edge down (changes the tone too) - worth checking its at full before each take. Compression can help to smooth out volume differences within reason but nothing can fix tonal variations so be carefull with that. That's why I always go direct (either Mesa Boogie Triaxis to Mesa poweramp to Palmer speaker simulator or a Rocktron Prophesy II) - I used to go direct from my amp via the 1/4" headphone socket but the sound wasnt the best - still more stable than I could manage by micing over days/weeks though.
Anyway, hope that helps - post back if you need any other help.