Discussion on:

Five tips for solving software problems

More often than not, the issues I receive are ones not reproducible and there's no obvious reason why it would fail.

Working backwards has saved me quite a few hours by examining how the code could've gotten to that point. Excellent suggestion!

I tend to work both ends when debugging but that requires a good knowledge of the code and some (hopefully intuitive) assumptions.

One thing can't be emphasised enough, give your intuition one chance. If it turns out to be wrong, stop guessing start finding.