Odours in waters used for various purposes, especially musty odours, is a worldwide problem. To establish an early warning system, the nuisance odour in an aquatic environment, the causative microorganisms and compounds, and the mechanism of odour in Lake Kasumigaura, the second biggest lake in Japan, were examined using in situ studies. In this study, the mechanism for the occurrence of musty odours was determined, and the level of musty odour was discriminated, based on 32 months of phytoplankton data from three sampling stations. The results indicate that the cell numbers of Phormidium tenue , Synedra and Ankistrodesmus were significantly related to the occurrence of 2-methlisoborneol (MIB) and geosmin. Furthermore, the odour occurrence for both MIB and geosmin was effectively predicted and validated using environmental factors as explanatory variables, using multiple linear regression and artificial neural network technology.