Abstract: Lower Es3 of Bijia fault block in Binnan Oilfield which contains as many as 58 oil subzones vertically belongs to the oilfield of multi-layer sandstone and strong heterogeneity reservoir.With the development of exploitation，conflicts between these layers are increasing.On the basis of comprehensive analysis of geologic features and development characteristics of the oilfield，we propose a criterion for quantitatively classifying development zones by hierarchical clustering subzones with K-means method，which proves effective in directing classification of development layers.In this paper，we presents the layer division criterion，based on which we divided Lower Es3 of Bijia fault block in Binnan Oilfield into three sets of development zones and also directed by economical theory and numerical simulation technology.Research practice shows that the presented layer division method using K-means is reliable，and the oil recovery is obviously enhanced.

Key words: complex fault block reservoir, production layer, K-means algorithm, numerical simulation, clustering analysis, Binnan Oilfield