Factors affecting the spatial variations of water quality of the Mazandaran coastal ecosystem was determined in summer 2012. For this purpose, water quality parameters (nutrients, temperature, conductivity, salinity, dissolved oxygen, pH, chlorophyll and turbidity) were evaluated along 4 transects (Amirabad, Babolsar, Noshahr and Ramsar) in the depths of 5, 10, 20 and 50m, using multivariate analysis methods. Based on the cluster analysis of data, the sampling sites could be classified into 5 distinct groups, including 35-50m water layer of station 50m in all transects, 0-5m layer in station 5m in all transects, Ramsar transect and the majority of surface and bottom layers of 3 transects including Noshahr, Babolsar and Amirabad. Based on discriminant analysis, 86.40% of the sampling sites were correctly classified. Factor analysis explained 87.53% of the total variance, the five principal components of which (viz. temperature, turbidity, nitrate, silica and ammonium) were considered as the most effective parameters on the spatial variation of water quality. This study suggests that the number of sampling locations can be reduced to two transects. Thermocline, transport of nutrients (specially phosphorus and ammonium) from rivers, sea floor, cage culture and the ctenophore, M. leidyi, were the most effective sources on spatial variations of water quality. Moreover, the multivariate statistical methods were found to be useful tools to recognize the spatial variations pattern along the Mazandaran coasts in summer.