Quantitative　analysis　of　rolling　bearing　fault　is　a　crucial　issue　in　the　condition　monitoring　of　rotating　machines.　Therefore,　a　quantitative　analysis　method　of　bearing　fault　is　presented　in　this　paper　based　on　the　unit　of　dynamic　mechanism　and　feature　extraction　algorithm　of　vibration　signal.　Based　on　this　method,　vibration　signals　for　different　sizes　of　rolling　bearing　fault　in　the　outer　race　were　simulated　using　a　dynamic　bearing　model　with　five　freedom　degrees.　Then,　the　quantitative　size　feature　was　extracted　from　the　vibration　signals　with　an　algorithm　of　fast　spectral　kurtosis　(FSK)　and　quantitative　diagnosis　of　rolling　bearing　fault　was　obtained.　The　effectiveness　of　the　proposed　method　was　verified　by　practical　defective　bearing　signals　of　different　defective　sizes.