Tension-type headache is the most prevalent form of headache. Nevertheless, the majority of current classification systems for this kind of headache rely on qualitative descriptions and clinical symptoms rather than practical rules, which are less distinct from migraineous features. The aim of this study was to introduce new diagnostic criteria for differentiation between CTTH and migraine patients using nonlinear analysis of surface EMG signals recorded from temporalis muscle. The subjects were eight patients with chronic tension-type headache, ten migraine patients without aura and eight healthy, age and sex matched volunteers. The focus of this study was on HOS analysis and examining the bispectral patterns and extracting features based on entropy concept for differentiation. The results showed that negentropy was the only feature, which could distinguish three studied groups. However, pairwise analysis indicated that NBE was capable of differentiating between migraine and CTTH patients. In addition, bispectral magnitude and specially phase diagrams, enabled us to visually differentiate studied groups.