The disassembly of complex capital goods is characterized by strong uncertainty regarding the product condition and possible damage patterns to be expected during a regeneration job. Due to the high value of complex capital goods, the disassembly process must be as gentle as possible and being adaptable to the varying und uncertain product's state. While methods based on data mining have already been successfully used to forecast capacity and material requirements, the determination of the product's or component's condition has become apparent in the recent past. Despite the rapid increase in sensor technology on capital goods such as aircraft engines and their use for condition monitoring due to countless interfering effects, it is only possible to react spontaneously to the product's condition. So far, we have concentrated on product condition-based prioritization of disassembly operations in a logistics-oriented sequencing strategy. In this article, we present an approach to predict disassembly process-planning parameters based on operational usage data using machine learning. With the prediction of disassembly forces and times, processes, tools and capacities can be efficiently planned. Thus, we can establish a componentfriendly disassembly process adaptable to varying product conditions. In this article, we show the successful validation on a replacement model of an aircraft engine.