Once upon a time we simply kept graphs and tables in our lab notebooks. We kept videos of behavior and the transcripts from those videos. For decades I kept huge binders of printed computer output. I kept those long hole-punched cards in brown cardboard boxes. They were waiting in case anyone wanted them. At most someone might want a tabulation of sex ratios, or nest failures, or diploid males. Usually these were laborious to discover.
Today we can do so much more. Your whole project may be based on data generated by someone else. Everything is on the computer. Everything should be open. You should take the time to document and make your work understandable and accessible.
Here is an article with some excellent advice on exactly how to do that, particularly for analysis rich work, called Ten Simple Rules for Reproducible Computational Research. Even if you do not consider your work to be particularly computational, there are excellent tips here. Happy scripting!
Thanks, Joan. Great post! Note that our paper on best practices in scientific computing, which covers some similar territory, has just been accepted by PLoS Bio, too: http://arxiv.org/abs/1210.0530