Online learning concerns analyzing a continuous stream of transient data and progressively updating the knowledge model without revisiting previously encountered examples. This paper proposes a new incremental fuzzy learning approach for online classification of data streams. It enables an existing fuzzy rule set to be efficiently updated based on a new training example without learning from scratch. The proposed algorithm can not only incrementally construct new fuzzy classification rules but also update the content with old rules to assimilate information from new data. The efficacy of the proposed incremental fuzzy learning method has been demonstrated in a set of simulation tests where the benchmark data sets were treated as data streams for learning.