This report describes an approach for extraction of product line requirements based on existing user documentation. The approach we describe in this report supports capturing of the information found in user documentation of legacy systems, e.g., user manuals, and the specification of this information in product line models, using, e.g., Use Cases. The main goal of the approach is expert load reduction; as we found out during our product line technology activities at IESE that the domain experts are a serious bottleneck during the product line scoping and modeling activities. The approach is based on a variability modeling approach that allows integrating arbitrary modeling formalisms to capture the extracted requirements in a product line model. We propose a conceptual model describing the transition from user documentation to product line artifacts describing common and variable elements of a product line model. The approach uses extraction patterns that allow an easy identification of text elements in user documents. The pattern are then used to create a significant part of the requirements specification and product line model, respectively and we present an extraction process that guides the user in using the pattern. In this report we present the approach itself and show the application of our approach for three case studies, one from the embedded/ mobile phone domain, two from the information system domain. The case studies have shown that the approach is applicable to information systems and embedded systems and that it produces relatively complete and correct results. With one of our case studies that was performed in real industrial settings we could not only show expert load reduction when applying the approach but also an overall time saving when applying the approach for product line scoping. A predecessor of this report with a different focus can be found in [JD03b].