This report describes the dosimetric analyses of a Phase I/II protocol, designed to examine the capabilities of an institutionally developed intensity-modulated radiotherapy (IMRT) system with respect to dose escalation. The protocol employed stringent dosimetric guidelines in the treatment of locally advanced head-and-neck squamous cell carcinomas (HNSCC) with radiotherapy alone using IMRT and the simultaneous integrated boost (SIB) technique. The first 14 patients enrolled on the protocol were included in this analysis. Escalating doses of 68.1 Gy (6 patients), 70.8 Gy (6 patients), and 73.8 Gy (2 patients) were delivered to the gross tumor volume (GTV) in 30 fractions. Simultaneously, constant dose coverage was given to the subclinical disease and the electively treated nodal regions, which received 60 Gy and 54 Gy, respectively, in all three cohorts. Parotid glands were spared to the degree possible without compromising target coverage. The following indices are reported for the GTV: (1) dose to specified percent volumes (e.g., D(98) and D(2)); (2) homogeneity index defined as the ratio (D(2) - D(98))/D(prescription); (3) biologically equivalent uniform dose (EUD); and (4) an index of conformality, PITV, defined as the ratio of volume enclosed within the prescribed isodose surface to the target volume. Treatments were planned and delivered with nine 6-MV photon beams using the multileaf collimator (MLC) "sliding window" technique. Mean doses to 98% of GTV were 68.4 Gy, 70.5 Gy, and 70.8 Gy, and average GTV dose homogeneity was 6.7%, 7.6%, and 8.8% for the three cohorts. The average doses to the parotid gland proximal to and distant from GTV were 41.3 Gy and 25.7 Gy, respectively. Dose distributions measured in phantom showed good agreement with calculations. Treatment of locally advanced HNSCC using SIB-IMRT as described is feasible. Treatment planning and delivery are safer and more efficient than with conventional three-dimensional processes. Predicted dose distributions can be accurately delivered with excellent conformality using dynamic MLC. At least one of the parotid glands can be adequately spared. Patient follow-up continues and will allow eventual quantitative correlation of delivered dose distributions with clinical outcomes.