Depending on what degree of robustness you want this, it would involve at  
a minimum writing your own printer driver, all the way up to hooking into  
the print spooler and interpreting various printer-level languages  
(Postscript, HP's PCL, and others) in order to extract the text.

As far as I know, none of those are things you can do directly from  
.NET/C#.

Of course, there is in fact already a built-in "text-to-file" printer  
driver you could use, as well as Microsoft's XPS printer driver, and even  
drivers that produce PDF.  If you can deal with requiring the printouts to  
be sent to a specific printer driver, I think those would be a much better  
solution.

If you need to capture all text sent to any printer, that's quite a bit  
more complicated (per hooking into the print spooler).  Again, using an  
existing solution might be better than trying to roll your own, since it's  
not like you're going to be implementing this in-process anyway (i.e.  
doesn't need to be integrated with whatever else you're doing).

In any case, I doubt this is the right newsgroup for trying to learn how  
to implement this.

Pete