Purpose: We designed and developed a daily QA software for patientsˈ treatment using dynamic MLC log files for IMRT and RapidArc. The log files of patientsˈ clinical treatments were also assessed retrospectively. Methods: The log files were sent to the server using our in‐house client software in a treatment console PC via a network, and then the errors of leaf position, leaf gap, and dose were analyzed using our in‐house server software in a PC. Dose errors in 2D plane were computed. In the retrospective data analysis, the log files of a total of 5871 treatment fields from 2010/11/02 to 2011/02/04 were collected from 2 Linac console PCs at the Cancer Center. The errors of leaf gap and dose between treatment methods were compared. Results: All of the log files were sent and analyzed normally. The results of analyzing the 5871 treatment fields were that the average values of the leaf gap errors [mm] were 0.42±0.26, 0.53±0.24, 0.42±0.26, and 0.47±0.84, and the average values of the dose errors [%] were −0.11±1.14, 0.32±1.28, −0.22±1.86, and −0.19±3.70, for non‐gated IMRT, gated IMRT, IMRT and RapidArc, respectively. Consecutive daily errors of the gaps and doses for RapidArc showed similar trends to those of IMRT. Since actual leaf gaps for RapidArc (21.29±11.72mm) were larger than those of IMRT (11.18±6.35mm), dose errors for RapidArc were decreased compared to IMRT. RapidArc had similar accuracy for dose distribution in terms of MLC movements. For gating, dose errors can arise from MLC gating signal hold‐on and hold‐off. Conclusions: The results of retrospective study could contribute to patient‐specific QA by providing a reference of the accuracy of MLC movement. Furthermore, the use of our system based on the retrospective analysis could document the accuracy of machine performance with regard to patient treatment.