Nowadays, there are two methods of MLC quality control. First method the patient specific quality assurance assumed that before every treatment start, plans are verified using for example electronic portal dosimetry. Second one the machine specific quality assurance assumed that MLC parameters like for example leaf position accuracy for static tests or picket fence for dynamic tests should be verified every established period of time. Both methods have specific advantages and disadvantages. The aim of this study was to introduce ARTISCAN software from AQUILAB for automated quality control for MLC machine specific quality assurance.The Artiscan software for automated quality control is using mathematical formulas to provide quantitative information about MLC parameters. So far in our department after MLC commissioning we only did patient specific quality assurance. To improve quality of treatment we have implemented this software. We made proper plans and realized them on our three Clinac 2300CD, one Unique and two TrueBeam equipped with MLC 120 Millennium and one TrueBeam equipped with MLC HD all from Varian. We analyzed all available parameters for static and dynamic tests. We manly focused on MLC leaf position static test and all picket fence dynamic tests. Plans have been realized for three months. Almost 3000 DICOM images from linacs were analyzed in ARTISCAN software.For four linacs the software for automated quality control showed almost 1.0 mm error for leaf position. We reported that to our service and service engineer calibrated MLC using standard procedure. After that calibration leaf position error decreased to maximum 0.5 mm. No tolerance deviations were observed for the rest of analyzed parameters.Both methods are complimentary but machine specific quality assurance approach can detect smaller difference and can improve accuracy of treatment. For patient specific quality assurance using electronic portal and gamma analysis we routinely use criteria 3% dose difference and 3 mm distance to agreement. These criteria will not detect such small differences in MLC working like 1 mm. However we still need more measurements to observe variability of parameters to adopt tolerance and to create levels of respond.