Purpose: Tumor volume after the lymph node involvement is one of the most important single prognostic factor in patients of head and neck cancers treated with radiotherapy. We have recently demonstrated that the hypoxic subvolume is more important than the total tumor volume. We therefore propose the hypothesis that the presence of visible necrosis might be an important factor for cure by radiotherapy in squamous cell cancers of the head and neck. Methods: A total of 51 patients with locally advanced inoperable (T3-4 or N2-3) squamous cell cancers of the head and neck (mean age 57 years, range 41-75 years) were prospectively investigated with regard to a possible impact of tumor volume. All patients received CT examination of the head and neck according to a standardized protocol (spiral CT, contrast enhancement after automatic injection), and the total tumor volume was calculated as the sum of volumes of all visible macroscopic tumor sites. Poorly perfused and necrotic areas (no contrast enhancement) within macroscopic tumor sites were also calculated. Patients were then treated with accelerated-hyperfractionated radiotherapy in about 6 weeks. Seventeen patients were treated with only radiation. Patients without contraindications to cisplatin chemotherapy received cisplatin chemotherapy or a combination of cisplatin and paclitaxel (N=34). The allocation of patients to certain treatment regimens was based on individual decisions in each case and not randomized. Results: In patients treated with radiation alone, 12/17 (71%) got recurrence whereas in patients treated with radiation plus cisplatin, only 14/34 (41%) recurred (P=0.05). The 2-year overall survival was for radiation alone versus radiation plus cisplatin 0% vs. 62% (P<0.0008). Tumors with smaller amount of necrosis (necrosis volume<4 cm3) had a good prognosis irrespective of type of treatment (radiation alone or radiation plus cisplatin). However, patients with tumors with a larger amount of necrosis (necrosis volume < 4 cm3) had a significantly better outcome if they were treated with radiation plus cisplatin as compared to patients treated with radiation alone. In a multi-variate analysis using a Cox-regression model the type of treatment (radiotherapy plus versus without cisplatin) was the only independent prognostic factor for event-free survival (P<0.03) in the whole group. Conclusions: In this non-randomized retrospective investigation with limited sample size, radiation plus cisplatin was superior to radiation alone. This resulted mainly from a higher efficacy of the radiochemotherapy regimen in patients with large and especially necrotic tumors. The prognostic and predictive impact of visible necrosis should be further evaluated.