Cryoablation has recently become a useful procedure for the treatment of prostatic and hepatic tumors, primarily because of advances in the ability to monitor visually the freezing process with ultrasound. Success of the procedure depends in large part on how well the ice front can be positioned to destroy pathologic tissue, while sparing healthy tissue. This study describes a cryogen delivery system that can be used in conjunction with magnetic resonance (MR) image-guided cryoablation, and an automatic control system that uses MR image guidance in a feedback loop to control the ice front trajectory. Edge-detected MR images are used to determine the current ice front location at each time interval, providing feed-back to an automatic control system that adjusts the flow of cryogen to the cryoprobe. Numerical simulations and experimental results demonstrate that an ice front with cylindrical symmetry can be accurately controlled using this MR image-guided feedback control scheme.