Purpose: IMRT (intensity-modulated radiotherapy) verification techniques are reviewed together with investigations demonstrating the intrinsic verification problems. Material and Methods: Different IMRT verification procedures for either class solutions or individual patients are demonstrated. Among the Latter are techniques Like fluence or three-dimensional (3-D) dose distribution verification within a transfer phantom. Different radiographic films and absolute dose probes are investigated for their suitability. Finally, Monte Carlo techniques (XVMC/VEF) are used for error detection and IMRT verification. Results: During introduction of clincal IMRT for head and neck (H&N) tumors, we concurrently applied fluence, relative, and absolute dose measurement. While fluence and relative dose are in rather good agreement with calculations, absolute dose is always low when compared to the TPS (TMS 6.1A, Nucletron B.V.) by 5-7%. This deviation seems to depend not on the number of segments, but can strongly depend on MLC misalignment. Further investigations have revealed the importance of a detailed commissioning of the TPS down to the small-fielld range using diamond or diode probes and its detailed verification. In addition, simple tests have shown that dose calculation approximations in the IMRT option of TMS are one major source of the dose deviation. XVMC/VEF does not use Such approximations. Conclusion: The procedure starts with a detailed TPS commissioning and verification process. Different verification methods are recommended during clinical IMRT implementation phase, in order to Locate sources of error. Later on, a minimal program could consist of a fluence or relative dose verification procedure with few films and absolute dose measurement, followed by an intensive MLC quality assurance (OA). Inverse Monte Carlo systems, Like IMCO++/IKO or Hyperion, seem to be able to reduce the effort.