Here, we introduce the protocol for small-scale and simple subcellular fractionation used in our recent publication (Taguchi et al., 2013), which uses homogenization by passing through needles and sucrose step-gradient. Subcellular fractionation is a very useful technique but usually a large number of cells are required. Because we needed subcellular fractionation of transiently-transfected cells, we developed a protocol for smaller numbers of cells. Our protocol for the subcellular fractionation is based on the protocol published by de Araújo and Huber (de Araujo et al., 2007), although substantial modifications have been made according to our experiences and information from personal communications. As optimal conditions seem to vary between cell lines, we advise to further modify the protocol to optimize for individual experiments. Our method is simple but sufficient for analysis of integral membrane proteins or proteins anchored to organelles by glycosylphosphatidylinositol or other lipid anchors, e.g. prion protein. However, proteins non-covalently attached to membranes or membrane proteins of organelles seem to be more prone to dissociation from the organelles during preparation and, if these proteins are the object of study, further modifications might be necessary. Unlike in a continuous gradient, where a protein of interest is scattered over a wide range, step-gradient fractionation is advantageous in detection of relatively small amounts of proteins from small-scale experiments, because it concentrates the protein of interest in one fraction, if an appropriate combination of sucrose concentrations is used.