The space before ( and after)  the song is done in the mastering stage not the mixing.
So it doesn't matter how much you leave in Sonar as long as there is at least a beat before any events or they might not trigger. So best to leave at least 1 blank measure. I leave 2. 
As said, more can be easily inserted at any time. 
 
You trim the beginnings and ending in a wave editor after exporting in the mastering stage. If you wish to do your own mastering I recommend using a wave editor which makes this task easy. 
 
I usually leave a touch over 1 second at the start of songs. About a beat and a half. 
The ending depends on the tempo so might be longer for a slow song. I just listen to the end and sort of envision where I'd like the next song to start.
These days there's no album order because songs will be added to peoples personal playlists so you don't now what your song will end up back to back with anyhow.
For CD's I fuss a little more with transitions and I tell my burning software to not add space. 

Thank you for your answers, 57Gregy and sock monkey. They both contain information that is going to come in very handy.
Cheers,
LJ

I think some people are confused eg Sharke. The OP is not about how much time that is needed before the start of the track on the final CD. I think he is referring to the amount of time to leave at the start in the multitrack session.
 
It is not wise to start the music on 00.00.00 in your multitrack session at all. The reason is that the moment you hit play the sequencer has just a few things to do before it can take off and often you will miss some very first notes etc..
 
It is better to leave two measures or so before the music starts in your multitrack session. Especially if you need to play midi notes early etc for synths that might have a slow attack sound etc..
 
If you suddenly get an idea for an extended intro then it should be easy in any DAW to pick up the whole session and move everything over and insert measures.
 
Regarding tracks on a CD I have found if you cut the music really tight for the track ID then some CD players will cue up just a little into the music and you might miss part of the very first transient. I have found from experience that leaving 100 mS of silence before the music starts will usually avoid that and the track will still cue quickly if required to. (eg a theatre situation)

It is not wise to start the music on 00.00.00 in your multitrack session at all. The reason is that the moment you hit play the sequencer has just a few things to do before it can take off and often you will miss some very first notes etc..
 
It is better to leave two measures or so before the music starts in your multitrack session. Especially if you need to play midi notes early etc for synths that might have a slow attack sound etc..
 
 

I'd like to hear from Cake on the first part of the above, as I've never heard any mention of not starting at 0:00.  Otherwise, why would that be a possibility on such a mature program?   I have to assume they've anticipated this scenario, since it is so easy to record at that point, and have allowed their internal *workings* to get up and running before the first note at 0:00 was played.  I have never lost any notes starting a project immediately at 0:00 either.
 
Regarding MIDI notes, that would be an anticipation, which happens a lot and I mentioned above about splitting clips and losing MIDI notes due to such anticipations.  If you know you're going to have to have a note start before 0:00, so the sound is ramped up and audible AT 0:00, it only makes sense to allow for needed space to due so.

mixmkr
 
I'd like to hear from Cake on the first part of the above, as I've never heard any mention of not starting at 0:00.  Otherwise, why would that be a possibility on such a mature program?

 
I'm with Jeff, and I would say that applies to any program. Granted, 99% of the time the sequencer will start up okay. But if you leave a blank measure, that becomes 100%. Remember, your computer is doing a ton of things that have nothing to with music, like scanning ports, keyboards, mice, turning USB off if you're not careful, etc. etc.
 
Think of it as the equivalent of a tape reel taking a fraction of a second to get up to speed.
 
As to space in a final product, I always like to leave a little bit of air at the beginning. I zoom waaaaaaaay in vertically, and cut just before "nothing" turns into "something."

Anderton
 
 
Think of it as the equivalent of a tape reel taking a fraction of a second to get up to speed.
 
 

I understand your analogy, but it isn't mechanical like a tape deck.  Granted, much of the time I might have anticipated notes before the first *hit* or downbeat of a song, so you do need that extra measure or so.  It seems odd to me that that availability to record right at [after] 0:00 is there. 
 
I'm talking unknowledged here...but I think we're talking about a fraction of a millisecond to *cue* up the sequencer/computer before 0:00 actually starts playback/record.   I don't know.  And I would also guess, if you had 30 tracks of virtual instruments {WITHOUT anticipated notes], all loaded up with plugs...it would narrow the gap between "computer ready" and "actual playback/record at 0:00".   Maybe to the point that yes....things would stumble.   (that 1% ??)

I'd love to hear Noel's take on this or whoever.   Remember...why is it OK 99% of the time?  Seems that if it is that frequency, the feature to start at 0:00 was taken into consideration by the Cake designers. 

Anderton
 
 
As to space in a final product, I always like to leave a little bit of air at the beginning. I zoom waaaaaaaay in vertically, and cut just before "nothing" turns into "something."

yes...especially if making a CD.  Most CD players as you know, will have a ramp up time, that will chop the first segments, if edited too close.

mixmkr
I'm talking unknowledged here...but I think we're talking about a fraction of a millisecond to *cue* up the sequencer/computer before 0:00 actually starts playback/record.   I don't know.  And I would also guess, if you had 30 tracks of virtual instruments {even without anticipated notes], all loaded up with plugs...it would narrow the gap between "computer ready" and "actual playback/record at 0:00".   Maybe to the point that yes....things would stumble.   (that 1% ??)

 
I don't think there's any way you NEED a measure, you probably only need a millisecond or two (even attack times of 1 ms are audible). However, by making it a measure, it's easier to move things around.
 
And it sure comes in handy if I want to add preverb 

Anderton
 
 
I don't think there's any way you NEED a measure, you probably only need a millisecond or two (even attack times of 1 ms are audible). However, by making it a measure, it's easier to move things around.
 
 

That statement seems to conflict with

Granted, 99% of the time the sequencer will start up okay.

 
And I'm not getting legalistic on you.  Sorry if it might appear that way...  {honest} ;-)
and I'm talking about 1/100th of a ms...   if that is even logically in the ballpark of this discussion.  Way beyond audio perception.

Soft synths , drums used to have problems missing the first note ,so it has been standard practice to leave a beat or 2.
also  say you are recording and   you are a fraction  ahead of the beat ,  there is no room for it .An acoustic strum is usually starts just before the beat .  To each his own, but i think there are many more reasons to leave space than not.