For treatments with dynamic intensity modulated radiotherapy (IMRT), the adjustment of multileaf collimator (MLC) parameters affecting both the optimization algorithm and dose distributions is crucial. The main parameters characterizing the MLC are the transmission (T) and the dosimetric leaf separation (DLS). The aim of this study is twofold: a methodology based on the 'sliding slit' test is proposed to determine (T, DLS) combinations inducing the best conformity between calculations and measurements. Secondly, the effects of the MLC adjustment on measured dose and on optimization are presented for different configurations as the chair test and for the patient dosimetric quality control (DQC). Tests were performed with a Varian 23EX linac operated at 20 MV and equipped with a 120 leaf Millenium dynamic collimator. The treatment planning system was CadPlan/Helios (version 6.3.6). Results demonstrated that the sliding width (SW) strongly depends on the (T, DLS) combinations, and the measured dose is a linear function of the SW. Different (T, DLS) combinations induced a good agreement between calculations and measurements. The influence of the MLC calibration was found to be particularly important on the 'sliding slit' test (11.8% for a gap change of 0.8 mm) but not so much on the chair test and on the DQC. To detect small variations in leaf adjustment and to ensure consistency between calculation and actual dose delivered to patients, a daily check called IMRT MU check is proposed.