is one of the important data mining techniques which discover clusters in many real-world data sets. Recent algorithms attempt to find clusters in subspaces of high dimensional data. Density based clustering algorithms uses grid structure for partitioning each dimensions into intervals (bins) which yields good computation and quality results on large databases. In this paper, we propose equal-frequency based (EFB) grid structure for efficient computation of clusters for high dimensional data sets. The computation is reduced by partitioning the bins with equal frequency bin method. The performance evaluation is done with data sets taken from UCI ML Repository. The result gives better quality clusters compared with other grid structures. (bins) where the width of the intervals may vary. Each unit has the same number of data points. The number of intervals is fixed by the user parameter. Also, density of an object in each unit is measured by the location of objects in the neighborhood within that same unit, instead of counting the number of data points in the unit. Taking the count of data points in each unit may include an outlier in the cluster. So in our approach, the distance between the data points in each unit are calculated and the units having minimum mean distance values greater than the threshold are taken as dense regions to form a cluster. The threshold is defined to be a fraction of the total number of records present in the unit.