I agree with Peter.

I tend to start with a very small section of code or the beginning of a program that is so simple I know it works. I test it and then start adding functionality, testing each new functionality as I add it. It's much easier to test incrementally and write-and-fix-as-you-go than to have some huge monster program and try to figure out where the bugs are hiding.

Erik's idea of testing certain sections of code and/or hardware by toggling certain pins is also a tried-and-true method of finding bugs. I often use a pin during development for the sole purpose of raising/lowering when I get to certain sections of code--which proves to me that it GOT to those sections of code. If it didn't get there then I don't even need to debug that section of code yet; I need to figure out why the program never got there.

Craig Steiner