The number of joint replacement surgeries and osteotomies has increased continuously over the last years and the ageing population is expected to further fuel this growth in the future. These surgical procedures require careful pre-operative planning based on exact alignment measurements, as well as post-operative assessment in order to evaluate the success of a treatment. In many clinics image acquisition workflow is already strongly optimized for patient throughput.However, performing orthopaedic measurements is still a time-consuming task which is subject to intra and inter-observer variability. With quantitative imaging in terms of an automatic segmentation of the bones and joints and subsequent derivation of relevant measurements it is possible to remove this bottleneck from the clinical routine. The major difficulties resulting from the challenging image data are (i) poor image quality resulting from the effort to keep radiation doses as low as possible and therefore demanding for robust features, (ii) a high degree of anatomical variability, especially caused by the development of bones during childhood and adolescence, and (iii) artificial objects such as implants, fixations, or clamps, in post-operative radiographs as a result of the treatment.This work presents a new method that addresses these challenges and automatically determines objective and reproducible measurements. The method has been evaluated on clinical data and compared against the readings of an experienced orthopaedist.