MRI-guided transurethral ultrasound therapy uses a linear array of transducer elements and active temperature feedback to create volumes of thermal coagulation shaped to predefined prostate geometries in 3-D. Numerical simulations have been used to determine robust feedback control algorithms, optimal transducer designs, effects of various tissue and imaging parameters, as well as to evaluate potential treatment accuracy and safety in patient-specific anatomical models. The goal of this work is to evaluate quantitatively the accuracy with which these numerical simulations predict the extent, shape and temperature pattern of 3-D heating produced in tissue-mimicking Zerdine* gel phantoms. Methods. Eleven experiments were performed in a 1.5T MRI scanner. Temperature feedback was used to control the rotation rate and ultrasound power of a transurethral device with five 3.5×5mm transducer elements. Heating patterns shaped to 23 and 11 cc human prostate geometries were generated using devices operating at 4.7 and 8.0 MHz, respectively, and 10W/cm2 surface acoustic intensity. Transducer surface velocity measurements were acquired using a vibrometer and used to calculate the resulting acoustic pressure distribution in gel. Temperature dynamics were determined according to a FDTD solution to Pennes' BHTE. Results. The numerical simulations predicted the extent and shape of the coagulation boundary produced in gel to within (mean± stdev [min, max]): 0.1± 0.4 [-1.4, 1.7] and 0.0± 0.3 [-1.0, 1.5] mm for the treatments at 4.7 and 8.0 MHz, respectively. The temperatures across all MRI thermometry images were predicted to within 10%, and the treatment time (∼20 min) to within 20%. The simulations showed excellent agreement in regions of sharp temperature gradients, near the transurethral and endorectal devices. Conclusion. Heating patterns predicted by the numerical simulations correspond closely to those produced experimentally in gel. This work quantifies the accuracy and demonstrates the validity of using numerical simulations to model MRI-guided transurethral ultrasound prostate therapy.