Natural sediments are a complex mixture of magnetic minerals with different origins and different geochemical history, each of which is called a magnetic component. Magnetic components practically never occur in isolated form, and their characterization using bulk magnetic measurements relies on the individuation of the systematic variation of some parameters within a large group of samples. These variations can be interpreted either as a mixing trend or as the result of natural processes, which affect the physical and chemical properties of the magnetic particles. An alternative approach is offered by the analysis of magnetization curves using model functions, which are supposed to represent the magnetic properties of individual components. The success of this approach relies on (1) the choice of model functions that can reproduce the natural properties of a component with sufficient accuracy by varying a minimum number of parameters and (2) on very precise and accurate measurements, which are necessary to overcome the extreme sensitivity of the method to noise. In this paper, the analysis of remanent magnetization curves proposed by Egli (2003) is applied to a large set of representative sediments from the most variable environments and to a set of artificial magnetite samples. Despite the variety of materials and natural processes involved in the formation of these sediments, seven groups of magnetic components with well-defined and consistent properties could be identified. It has been found that both lacustrine and marine sediments contain two magnetically distinct groups of magnetosomes, which react differently to changes of the redox potential. The effects of some natural processes, such as weathering, reductive dissolution and transport could be observed on the individual components.