Magnetic sensors and nonlinear models of magnetic fields (as functions of distance) can be used for position estimation in a number of real-world applications. However, disturbances from foreign ferromagnetic objects and from other spurious magnetic fields can cause significant errors in position estimates. This paper develops a disturbance compensation system based on redundant magnetic sensors and use of adaptive disturbance estimation algorithms. The developed disturbance compensation system is used for position estimation in a pneumatic cylinder application. Experimental results demonstrate that sub-mm accuracies can be obtained in position estimates, even in the presence of foreign ferromagnetic objects. The developed position estimation system can be used in a large number of industrial and mobile applications for piston position estimation tasks.