Magnetic resonance (MR) tagging is a method for visualizing cardiac motion in MR images by inducing artificial, high‐contrast features that move with the tissue. To enable quantitative analysis of tagged MR images, a method for tracking the movement of tag features is required. In this article, we present a new method for tracking the deformation of planar tags applied to the left ventricle of the heart. The method generates a continuous three‐dimensional reconstruction of deformed tag planes using optimal estimation theory. Tag plane movement is modeled as a time update process, where each update consists of a quadratic component that describes the bulk motion of the tag planes plus a stochastic component that characterizes the fine‐scale variation in the motion. The model is used in conjunction with a novel, recursive algorithm that efficiently estimates the deformation of tag planes from time‐series data. Results are shown using both simulated and actual MR images. © 1999 John Wiley & Sons, Inc. Int J Imaging Syst Technol, 10, 128–142, 1999