16 bit 0r 24 bit?

16-bit works well enough as a format for a playback system like CDs, though DVDs and the "high definition" CDs (if they ever take off) use 24-bit. But playback of a finished, mixed and mastered track isn't the same thing as recording and processing the audio in the first place.

The advantages of 16-bit over 24-bit for recording are that you'll save disk space as 16-bit files are smaller than 24-bit ones. There's a bit less system (disk/cpu) load involved as well, but usually not enough to bother a modern PC.
 
That's about it.

The advantage of 24-bit, from the point of view of recording, processing and mixing audio, is that 24-bit has a much greater  dynamic range than 16-bit (the difference between the quietest and loudest sound it can store or reproduce). 24-bit also has a better signal-to-noise ratio than 16-bit.

To put some numbers to it, 16-bit has a dynamic range of 96dB (less than the human ear). 24-bit has a theoretical dynamic range of 144dB, though in the real world the limitations of the electronics mean something like 124dB is more likely. Either way, that's a huge difference.
 
What this means in practice is that in 24-bit you can set your levels lower while recording and still capture a good dynamic range.
 
This matters because digital recording is very unforgiving about levels. Go "over" the number of available bits and you get digital clipping, where the sound breaks up instantly into pure noise, which can't usually/easily be sorted out afterwards other than by doing another take at lower levels.

So to avoid clipping you must make sure you don't go "into the red" (over 0dB) on Sonar's track meter while recording to that track. So several dB have to be left "spare" while recording to make sure any brief transient peaks don't clip. 

In 16-bit, keeping the meters in the safety zone, say below -6 or -12dB (or more depending on what/who you're recording) means you've given away quite a chunk of the available dynamic range.

You can allow the same or more "spare headroom" in 24-bit and still have greater dynamic range than the maximum 16-bit is capable of, even if you could track using all those 16 bits - which you can't because of the risk of clipping.

You also need to be sure you don't overload the convertors in the interface and make them clip and, again, 24-bits gives you more room to start with.

Higher bit depths also handle processing better than lower bit depths.
 
Finally, having the individual tracks recorded at levels somewhat lower than 0dB makes keeping plugins, the master bus and other busses out of the red easier while mixing.

In general therefore, I suggest recording, mixing etc. at 24-bit  then to put the end result on a CD convert it to 16-bit as a final step.

As for the 44.1KHz sample rate, opinions vary. Search the Sonar forums and you'll find loads of discussions about sample rates. Me, I record at 48KHz, but probably as much out of habit as any other reason (it was the best the DAW I had about 10 years ago could cope with and I've just kind of stuck with it ever since).

One practical advantage of using higher sample rates is that they give you lower latency from the interface (the time it takes for audio to get in and/or out of the DAW, which matters for playing software synths for example), but at the cost of more PC processing power being needed and more disk activity (higher rates=bigger files).

How well your pc can handle the data at higher sample rates is a big factor in all of this. Often it's a matter of trying things out and seeing what happens, tweaking then trying again. DAWs are like that.

I hope all that makes sense. If it doesn't that's my fault not yours, so don't be afraid to tell me so. "If in doubt, ask" is always the wisest course to follow :-)

Nice explanation tlw