We describe our experiences with four separate conversions from paper documents into hypertext and discuss the lessons we have learned. The paper document's organization affects the ease with which it can be converted and the appropriateness of the resu lting hypertext. The form of the paper document's machine-readable `markup' description affects the ability to transform the structure automatically. Designing the link structures that tie together the parts of the hypertext takes special care in automa ting, as badly-designed and incorrectly-formed links destroy the integrity of the hypertext. Overall, each of the conversions followed the same basic methodology, providing the handle for the development of `power tools' that can be applied to simplify s ubsequent conversions.