Microarray experiments usually generate data sets with multiple missing value due to several reasons. In the paper a robust method has been proposed to estimate the missing value of microarray experimental data. Missing values are imputed using fuzzy similarity measure by identifying the genes having similar characteristics to that of the gene with missing values. In this approach, biological knowledge of the gene is extracted using fuzzy relation and based on that knowledge, missing value is predicted and optimized. The estimation accuracy of the proposed method is compared with the existing K-nearest neighbour (KNN) based missing value imputing method. The result demonstrates that the proposed method outperforms the KNN based method.