In medical field, radiologists are more interested in region of interest (ROI) rather than whole image. ROI is a subpart of the image that contains very important information related to the diagnosis. In addition, ROI size has been known to influence the sensitivity and specificity of the classification. Many researchers in the past have used ROI for texture analysis of fatty and dense mammogram. Since a ROI is used as ‘representative’ of the image and all further computations and diagnosis depends upon the ROI, therefore, it is very crucial to select an appropriate image area as ROI. In this work, experiments have been conducted to find the appropriate size of ROI for breast density classification. Comparison of different ROI sizes 50 × 50, 75 × 75, 100 × 100, 125 × 125, 150 × 150, 175 × 175, 200 × 200, 225 × 225, 250 × 250, 275 × 275 and 300 × 300 pixels for breast density classification has been done. Different texture models have been used to extract features from these ROIs to have reliable analysis for appropriate size of the ROI. The effect of ROI size is evaluated in terms of the performance of differentiating between the fatty and dense breast tissue. For carrying out the experiments standard publicly available mini-MIAS database has been used. After the analytical study, it has been observed that a square shaped ROI having size of 200 × 200 pixels is optimal and it should be taken from the central breast region immediately behind the nipple, as this region is the densest region of the breast excluding pectoral muscle. The optimal ROI size also reduces the computational cost in extracting the texture features from the small sized ROI. The experimental results encourage the use of 200 × 200 pixels ROI for the classification of breast density.