Spatial colocalization of fluorescently labeled proteins can reveal valuable information about proteinprotein interactions. Compared to qualitative visual interpretation of dual color images, quantitative colocalization analysis (QCA) provides more objective evaluations to the degree of colocalization. However, the finite resolution power of microscopes and the spatial patterns of intracellular structures may compromise the reliability of many classical QCA methods. In this paper, we discuss the strength and weakness of some mostly used QCA methods. By studying their applications on computer-simulated images and biological images, we show that classical pixel intensity based QCA methods are often vulnerable to coincidental overlapping among resolution elements (resel) distributions and thus not suitable to images with high molecular density or with low resolution. Also, many QCA methods can mistakenly regard long range correlation as colocalization due to protein localization in intracellular structures. The newly developed protein-protein index (PPI) approach is able to reduce the influence from resel overlapping and spatial intracellular pattern compared to previous methods, significantly improving the reliability of QCA.