One aim of adaptive radiotherapy (ART) is the observation of organ motion followed by a subsequent adaptation of the treatment plan. One way of achieving this goal is a kV x-ray source mounted at a linear accelerator in combination with a flat-panel imager. Two imaging hardware configurations were evaluated for their potential for online tracking and the subsequent correction of organ motion by using fluoroscopic images: x-ray tube positioned with (A) 90 degrees and (B) 180 degrees offset to the MV beam. For one lung case two IMRT plans with five coplanar beams and the table positioned at 0 degrees were optimized for two multileaf collimators (MLCs) with 10 mm and 2.75 mm leaf width. Respiratory motion, modelled by rigid transformation in the lungs, was investigated for different amplitudes. The 3D dose distributions for different cases (no movement, uncorrected movement, correction for the movement perpendicular to the respective kV beam) were evaluated with the help of dose volume histograms (DVHs) and a modified conformity (Baltas et al 1998 Int. J. Radiat. Oncol. Biol. Phys. 40 515-24) and coverage index using the 90% isodose. For the corrected treatment plans the influence of the observed displacement vector caused by organ movement was accounted for by a respective displacement of the target point. For the simulated movement with a small amplitude (3 mm) in the anterior-posterior (AP) direction the dose distributions resulting from the correction of the displacement vector using imaging system A or B showed similar results for both systems and were in good agreement with the dose distribution of the static (not moving) patient. Increasing the amplitude in the AP direction to 6 mm or even 9 mm leads for both amplitudes and both MLCs to almost the same conformity and coverage index as the static dose distribution if imaging system B is used for the online correction. For the dose distribution obtained with correction based on imaging system A the deviation between the optimal and the corrected dose distribution is increasing with increasing amplitude. For the MLC with the smaller leaf width the difference between the optimal and the corrected dose distributions is always significantly larger than for the less conformal dose distributions created by the MLC with the 10 mm leaves. These results can be explained by the fact that system A cannot observe movement in the AP-LR plane perpendicular to the MV beam and therefore cannot correct for these movements whereas system B only fails to observe the motion in the beam direction which for photon irradiation has less impact on the dose distribution.