IMRT plans are usually verified by phantom measurements: dose distributions are measured using film and the absolute dose using an ionization chamber. The measured and calculated doses are compared and planned MUs are modified if necessary. To achieve a conformal dose distribution, IMRT fields are composed of small subfields, or "beamlets." The size of beamlets is on the order of 1 x 1 cm2. Therefore, small chambers with sensitive volumes < or = 0.1 cm3 are generally used for absolute dose verification. A dosimetry system consisting of an electrometer, an ion chamber, and connecting cables may exhibit charge leakage. Since chamber sensitivity is proportional to volume, the effect of leakage on the measured charge is relatively greater for small chambers. Furthermore, the charge contribution from beamlets located at significant distances from the point of measurement may be below the small chambers threshold and hence not detected. On the other hand, large (0.6 cm3) chambers used for the dosimetry of conventional external fields are quite sensitive. Since these chambers are long, the electron fluence through them may not be uniform ("temporal" uniformity may not exist in the chamber volume). However, the cumulative, or "spatial" fluence distribution (as indicated by calculated IMRT dose distribution) may become uniform at the chamber location when the delivery of all IMRT fields is completed. Under the condition of "spatial" fluence uniformity, the charge collected by the large chamber may accurately represent the absolute dose delivered by IMRT to the point of measurement. In this work, 0.6, 0.125, and 0.009 cm3 chambers were used for the absolute dose verification for tomographic and step-and-shoot IMRT plans. With the largest, 0.6 cm3 chamber, the measured dose was equal to calculated within 0.5%, when no leakage corrections were made. Without leakage corrections, the error of measurement with a 0.125 cm3 chamber was 2.6% (tomographic IMRT) and 1.5% (step-and-shoot IMRT). When doses measured by a 0.125 cm3 chamber were corrected for leakage, the difference between the calculated and measured doses reduced to 0.5%. Leakage corrected doses obtained with the 0.009 cm3 chamber were within 1.5%-1.7% of calculated doses. Without leakage corrections, the measurement error was 16% (tomographic IMRT) and 7% (step-and-shoot IMRT).