Nowadays most people can get enough energy to maintain one-day activity, while few people know whether they eat healthily or not. It is quite important to analyze nutritional facts for foods eaten for those who are losing weight or suffering chronic diseases such as diabetes. This paper proposes a novel type-2 fuzzy ontology, including a type-2 fuzzy food ontology and a type-2 fuzzy markup language (FML)-based ontology, for diet assessment. In addition, we also present a type-2 FML (FML2) to describe the type-2 fuzzy ontology and the FML2-based diet assessment agent, including a type-2 knowledge engine, a type-2 fuzzy inference engine, a diet assessment engine, and a semantic analysis engine. In the proposed approach, first, the nutrition facts of various kinds of food are collected from the Internet and the convenience stores. Next, the domain experts construct the type-2 fuzzy ontology, and then the involved subjects are requested to input the different food eaten. Finally, the proposed FML2-based diet assessment agent displays the diet assessment of the food eaten based on the constructed type-2 fuzzy ontology. Using the generated semantic analysis, people can obtain health information about what they eat, which can lead to a healthy lifestyle and healthy diet. Experimental results show that the proposed approach works effectively where the proposed system can provide a diet health status, which can act as a reference to promote healthy living.