Dynamic programming (DP) is a mathematical technique for making optimal decisions on the sequencing of interrelated problems. It has been used widely to detect borders in magnetic resonance images (MRI). MRI is noninvasive and generates clear images; however, it is impractical for manual measurement of the huge number of images generated by dynamic organs such as those of the cardiovascular system. A fast and effective algorithm is essential for on-line implementation of MRI-based computer aided measurement and diagnosis. In this paper, a branch-and-bound dynamic programming technique is applied to detect the endocardial borders of the left ventricular. The proposed branch-and-bound method drastically reduces the computational time required in conventional exhaustive search methods. Statistical tests are conducted to verify the CPU time performance of the branch-and-bound technique in comparison to the conventional exhaustive search method.