In developing software, one of the important role is non-functional requirements. They are often forgotten by the developer, so that it causes adverse effects. For this reason, a non-functional requirement classification method is needed to make it easier for software developers to classify non-functional requirements from requirement document. Most of studies in this area are focusing on single label classification. However, a non-functional requirement sentence can be classified to more than one class. Therefore, a classification technique that support multi-labels was needed. This research proposes a model for classifying non-functional requirements into quality aspects based on ISO/IEe 25010. It uses Fuzzy Similarity KNN (FSKNN) for building the sentence-based classification model. Six difference dataset which is contain 1432 sentence is used in the test. Best accuracy, precision and recall value from classification using FSKNN respectively is 42.8%, 68.1%, and 55.9%.