We investigated the role of collagen in the magnetization transfer (MT) effect in contrast to other macromolecules. By means of phantoms made of collagen, chondroitin sulfate (CS) and albumin, MR parameters have been optimized in order to reduce the acquisition time and improve the sensitivity, as well as to minimize the contributions from CS and albumin to the MT induced signal attenuation. The same method was used to study cartilage ex vivo (bovine articular and nasal cartilage plugs) and in vivo (goat knee femoral chondyle). In phantom samples, the MT signal attenuation depended on the collagen concentration while contributions from the other macromolecules were found to be minimal. In average, analysis of MT images revealed a approximately 25%, approximately 35% and approximately 30% signal attenuation in 10% w/v type I collagen gels, cartilage plugs, and cartilage from the weight-bearing areas of the goat knee, respectively. Biochemical data revealed that treatment of cartilage plugs with bacterial collagenase led to collagen depletion and correspondingly to a decrease of the MT response. In contrast, trypsin-induced proteoglycan loss in cartilage plugs did not alter the MT effect. A significant correlation was observed between the collagen content in these plugs and their respective MT ratios and the rate constant k for the exchange process bound versus free water. Finally, data obtained from in vivo MT measurement of the goat knee demonstrated that intra-articular injection of papain might not only cause degradation of proteoglycans but also a change in collagen integrity in a dose-dependent manner. We conclude that in vivo measurement of MT ratios gives quantitative and qualitative information on the collagen status and may be applied for the routine evaluation of normal and abnormal articular cartilage.