Intensity modulated radiation therapy is today performed with such a conformity to the internal target volume in the patient that even a small misalignment between the incident beams and the target can dramatically reduce the effectiveness of the treatment. Consequently, there is a need for a measure that could quantify the accuracy of a delivered treatment in terms of expected clinical outcome. To evaluate such a measure, a cervix cancer was selected as the tumor site on the grounds that the involved organs at risk, mainly the bladder and the rectum, are very close to the tumor and partly located inside the internal target volume. In this work, a solid phantom simulating the pelvic anatomy was fabricated. A treatment plan delivering an IMRT dose distribution was designed using the anatomy of the phantom. The phantom, with a film positioned into it, was irradiated. The dose distribution delivered was derived from the film and compared with the one of the treatment plan. The expected complications for the delivered therapy are higher for the bladder (3.0%), lower for the rectum (-7.4%) and unchanged for the small bowel with an overall risk, PI deviation of -4.4%. For the target volumes involved, the gross tumor control is a little lower (-0.9%), but significant for the the control probability for the lymph nodes and the ITV (-10.8% and -11.3%, respectively). It is shown, that the physical comparison between the planned and delivered dose distributions do not generally express their real difference in treatment effectiveness. It is demonstrated how small inaccuracies in dose delivery can considerably deteriorate a IMRT treatment plan. The clinicians need to know how much the expected complication and control rates will increase and decrease respectively because of uncertainties in dose delivery. In IMRT delivery, the reliability of the patient setup procedure becomes critical for the effectiveness of the treatment.