As　one　of　the　key　components　in　rotating　machinery,　rolling　bearings　can　affect　the　running　state　of　equipment　and　even　cause　huge　damage.　Therefore,　various　methods　have　been　proposed　to　improve　the　processing　of　bearing　fault　signals.　Among　them,　spectral　amplitude　modulation　is　an　empirical　and　automated　method　that　can　extract　the　fault　characteristics　of　rolling　bearings　effectively.　However,　the　characteristic　frequencies　can　be　no　longer　evident　or　even　over-whelmed　due　to　strong　background　noise.　In　this　paper,　a　time-frequency　spectral　amplitude　modulation　(TFSAM)　method　is　proposed.　The　proposed　method　obtains　the　amplitude　in　the　time-frequency　domain　using　short-time　Fourier　transform　and　modifies　them　with　different　weights.　Thus,　more　accurate　and　detailed　information　about　the　amplitude　can　be　calculated.　Meanwhile,　an　indicator　is　proposed　to　select　the　optimal　weight　automatically,　after　which　more　obvious　characteristic　frequencies　can　be　spotted.　The　effectiveness　of　the　method　is　verified　by　a　simulated　signal　and　applied　to　rolling　bearing　fault　diagnosis　of　outer　ring,　inner　ring　and　compound　faults　respectively.　The　results　indicate　its　accuracy　and　effectivity　in　identifying　fault　information.　Comparisons　with　other　commonly　used　techniques　verify　the　advantages　of　the　proposed　method.