Objective. An MRI-based technique for non-invasive assessment of the quantitative distribution of articular cartilage in the knee-joint was to be developed, and its accuracy and reproducibility tested.Design. Three cadaveric specimens and one patient were studied and MRI measurements compared with anatomical sections or arthroscopy.Background. Data on articular cartilage thickness is needed for the design of computer models, determination of cartilage material properties from arthroscopy and staging of osteoarthrosis.Methods. The knees were imaged using strongly T2-weighted spin-echo and FISP-3D sequences. After digital subtraction and automatic segmentation, three-dimensional reconstruction of the cartilages was performed. Surface areas, volumes and the mean cartilage thickness were calculated, and the regional distribution displayed after trigonometric correction.Results. The difference between MRI volumes and those obtained from the sections ranged from 4 to 21% with a reproducibility of ±4 to ±12% after repositioning. The thickness maps obtained with MRI were very similar to those from the sections. In the patient, a full-thickness defect demonstrated with MRI was verified by arthroscopy.Conclusions. Using the technique presented, the quantitative distribution of knee-joint cartilage may be analysed non-invasively, accurately, and in a very time-effective manner, in cadavers and in living subjects.RelevanceTo date there exists no accepted method for the accurate, fast and non-invasive assessment of articular cartilage thickness. Such a technique is, however, very helpful for generating computer models of diarthrodial joints, determination of cartilage material properties during arthroscopy, staging of joint disease, and objective control of chondroprotective treatment.