Although intensity modulated radiation therapy (IMRT) offers the opportunity to achieve a better compromise between target volume coverage and sparing organs at risk, its routine clinical implementation is partially held back by the excessive time required to perform pre-treatment quality control. The purpose of our study is to evaluate and then optimise the pre-treatment quality control in IMRT, using Statistical Process Control (SPC). SPC is a method currently used in industry for controlling and improving the quality of a process through statistical analysis. SPC shows that, instead of monitoring individual results, in order to control a process it is necessary to simultaneously monitor the location and dispersion of subgroups of data. This practice constitutes an important change to current methodology. A retrospective statistical analysis showed that the distribution representing the differences between the measured and the calculated dose could be approximated by a normal distribution. This implies that the treatment delivery process is not subject to assignable causes, but to many factors that occur randomly. We then retrospectively evaluated the interest of setting a SPC. We found that using control charts during quality controls would have allowed us to detect genuine deviations, such as multileaf collimator calibration, more quickly. The control charts also confirmed that our process is only subject to causes of random variability, which means it is statistically controlled. The next step will be to try to reduce the causes of random variability. Finally, the capability analysis estimated our process performance and confirmed the descriptive statistics results: our process is not centred on the target value (0% difference between the calculated and the measured dose) and data dispersion is too large. Indeed, 4,3% of the results were outside the clinical thresholds. The next step will be to identify all the factors affecting our process by looking at the design of experiments, and then to take action when it is possible