Objective: To evaluate the prediction accuracy of three intraocular lens power calculation formulas (SRK/T, Haigis and Barrett Universal Ⅱ) for postoperative refractive power using a Pentacam AXL panoramic biometer in cataract patients with ultra-high myopia. Methods: This was a prospective clinical study. Thirty-two cataract patients (56 eyes) with ultra-high myopia [axial length (AL) ≥28 mm] who underwent cataract surgery in the Ophthalmology Department of the First Affiliated Hospital of Chongqing Medical University from October 2017 to March 2018 were selected. The Pentacam AXL panoramic biometer was used to measure the anterior and posterior corneal surface curvature, anterior chamber depth and AL. An intraocular lens (IOL) with appropriate residual refractive power was individually selected for each eye, and the postoperative residual refractive powers of the three formulas were obtained. The refractive powers of the patients at three months after surgery were measured. Then the difference between the preoperative residual refractive power and postoperative refractive power, which was called the mean refractive error, the absolute value of the difference, which was called the mean absolute refractive error (MAE), were calculated. The quantitative data of non-normal distributions were expressed by the median and percentile (P25, P75), and the measurement data were tested by a Mann-Whitney U test or chi-square test. Linear regression analysis was used to evaluate the correlation between AL and MAE. Results: The median and percentile (P25, P75) of the mean refractive errors of the SRK/T, Haigis and Barrett Universal Ⅱ formula were -0.18(-0.53, 0.23), -0.18(-0.54, 0.09), -0.11(-0.49, 0.15), respectively. The mean refractive error of the Barrett Universal Ⅱ formula was significantly lower than that of SRK/T formula (Z=-2.517, P=0.012). However, the Barrett Universal Ⅱ formula agreed with the Haigis formula, the Haigis formula agreed with the SRK/T formula, and the differences in mean refractive error were not statistically significance (P>0.05). The median and percentile (P25, P75) of MAE of the SRK/T, Haigis and Barrett Universal Ⅱ formulas were 0.40(0.20, 0.61), 0.32(0.14, 0.64), 0.27(0.12, 0.55), respectively. And there were no significant differences among the Barrett Universal Ⅱ, Haigis and SRK/T formulas (P>0.05). Moreover, the MAE is related to AL. In the SRK/T formula, an increase of 1 mm in axial length led to an increase of 0.23 D in MAE; in the Haigis formula, an axial length increase by 1 mm led to a MAE increase of 0.04 D; but in the Barrett Universal Ⅱ formula, an axial length increase of 1 mm led to an absolute error reduction of 0.01 D. Conclusions: The three IOL power calculation formulas are relatively accurate for postoperative refractive error in cataract patients with ultra-high myopia. The Barrett Universal Ⅱ formula has more advantages than the SRK/T formula and Haigis formula.