AIM: To evaluate the accuracy and predictability of ray tracing-assisted intraocular lens (IOL) calculation function in Sirius internal software and further improve the accuracy by optimizing the calculation of predicted lens position (PLP). METHODS: This retrospective study recruited 52 eyes of 49 patients. All of the cases with cataract had undergone phacoemulsification combined with IOL implantation. SRK-T, Haigis formula, and Sirius ray-tracing method were all used for each eye's IOL calculation. The mean absolute value of prediction error (prediction error=predicted refraction- postoperative refraction) was defined as mean absolute prediction error (MAPE) and was determined for each method. Calculation of PLP was optimized by effective lens position (ELP). Optimized PLP was entered to Sirius internal software again to verify whether the method was improved. RESULTS: Compared with SRK-T and Haigis formulas, less accuracy was shown in Sirius ray-tracing method (P=0.001). The ELP of the IOL moved forward compared to PLP (P<0.001). The MAPE of the ELP-inputted Sirius ray-tracing method was reduced. ELP and PLP were well correlated. Taking ELP as y and PLP given by Sirius soft as x, a linear regression formula y=0.1637x+3.1741 was concluded (R²=0.1066, P=0.018). It was shown that the optimized Sirius ray-tracing method (optimized PLP entered), compared with SRK-T and Haigis formulas, worked with the same accuracy (P=0.038). CONCLUSION: The original Sirius ray tracing method is not satisfactory enough. However, in normal eyes, the optimized Sirius ray-tracing method in IOL calculation was as accurate as SRK-T and Haigis formulas.