To examine the potential of dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) for differential diagnosis of head and neck cancer. DCE-MRI was performed in 26 patients with untreated squamous cell carcinoma (SCC), 28 undifferentiated carcinoma (UD) and 8 lymphoma. DCE-MRI was analyzed with the pharmacokinetic model proposed by Tofts and Kermode to produce the three DCE parameters: k(trans), v(e) and v(p). Areas under the curve (AUC) at the initial 60 and 90s (AUC60 and AUC90) were also recorded. Histogram analysis was conducted to obtain the mean, 25%, 50%, 75% and 95% percentile values and the Kruskal-Wallis test was used to compare the DCE parameters between the three groups of cancer. k(trans), AUC60 and AUC90 showed significant differences (p<0.01) between UD/SCC and UD/lymphoma, but not between SCC/lymphoma. The mean AUC90 demonstrated the highest accuracy of 78% (sensitivity of 68% and specificity of 88%) for distinguishing UD and SCC, and the 75% percentile AUC90 provided the highest accuracy of 97% (sensitivity of 100% and specificity of 88.5%) for distinguishing UD and lymphoma. There are significant differences in the DCE parameters which show the potential for distinguishing UD from SCC or lymphoma.