Aim of this report was to summarise clinical experience in the field of pre-treatment dosimetric verification of intensity-modulated photon beams (IRMT). From May 2001 to July 2003, 50 patients were irradiated according to IMRT techniques with 6 MV photon beams produced by a Varian Clinac equipped with a 80 leaves multileaf collimator. Dose plans were computed using commercial treatment planning systems, Nucletron Helax-TMS for static cases and Varian Eclipse-Helios for dynamic cases. Pre-treatment dosimetric verification was carried out on a field-per-field basis measuring 2D absolute dose distributions in solid water at 10 cm depth using films or an electronic portal imaging device (EPID). Verification measurements were compared with expected dose maps, and differences were evaluated by means of both a point-to-point analysis and the Gamma Index. Irradiated target volumes (30 head and neck, 8 breast, 12 other patients) ranged from 111 to 2121 cm3 with a mean of 652 +/- 378 cm3. Twenty-nine dose plans were delivered with dynamic technique and 44 with static technique. On average, 5.9 +/- 1.3 fields were applied per plan, with 12.1 +/- 1.6 segments per field in the static mode. Averaging over the whole number of fields we obtained a mean difference (on a pixel-by-pixel basis and per 100 MU delivered) of -0.22 +/- 0.64 cGy between calculation and measurement, with a standard deviation of 1.93 +/- 0.65 cGy. The mean value for the Gamma Index evaluation was 0.47 +/- 0.10, with a mean standard deviation of 0.35 +/- 0.17. The fraction of pixels lying inside the field and showing a gamma index larger than 1 was 5.7% for the triplet Eclipse-film-dynamic delivery and 9.9% for the triplet Helax-TMS-EPID-static delivery. The employed IMRT treatments proved that this technique is feasible and dosimetrically accurate. Treatment verification stability and dosimetric analysis of treated plans are highly satisfactory and allow the safe introduction of this modality in the spectrum of techniques offered to a large class of patients.