Intensity-modulated radiotherapy treatment (IMRT) has shown a better protection for organs at risk than conventional radiotherapy treatment in most cases. However, existing systems based on anytime optimizing schemes such as gradient search or simulated annealing are limited by the combinatorial complexity of the underlying search problem. Therefore, an inverse planning system entirely based on linear programming has been developed and implemented. Our system incorporates a number of improvements to achieve practicality of the linear programming approach. The main benefits of this system are (1) very short response time and (2) completeness properties of the optimization process. Our system MIPART (Munich Inverse Planning and Radiotherapy Treatment) has been integrated into a clinical environment and has been evaluated by oncologists and physicists.