In a real life environment, spatial data is highly skewed. In general, there are two kinds of skews in spatial data. One is the placement skew and the other is the area skew. This paper introduces methods and the complexity of estimating the result sizes of the multi-way join for the area skewed spatial data. Especially, this paper describes the number and sort of the statistics which the optimizer should keep in order to calculate the multi-way join result size. (C) 2000 Elsevier Science B.V. All rights reserved.