How to begin?

I finished the Tokenizer, now that I have all the tokens available for a particular program I am confused to implement the CompilationEngine as I don't know how to begin!
Since a program can start with any token( although most of them begin with class keyword), it can be a class, function or variable declaration etc How should I go about it?
Should I take each token and check for what type of token is it? and call for a subroutine?

[WBahn]

If you look at the language description, I think you will find that ALL files MUST begin with a 'class' token. I'm pretty sure that's true -- it's been a while since I've looked at it and I don't have the text handy.