Rapid　detection　and　classification　of　cancer　cells　with　label-free　and　non-destructive　methods　are　helpful　for　rapid　screening　of　cancer　patients　in　clinical　settings.　Here,　surface-enhanced　Raman　scattering　(SERS)　was　used　for　rapid,　unlabeled,　and　non-destructive　detection　of　seven　different　cell　types,　including　human　cancer　cells　and　non-tumorous　cells.　Au　nanoparticles　were　used　as　enhanced　substrates　and　directly　added　to　cell　surfaces.　The　single　cellular　SERS　signals　could　be　easily　and　stably　collected　in　several　minutes,　and　the　cells　maintained　structural　integrity　over　one　hour.　Different　types　of　cells　had　unique　Raman　phenotypes.　By　applying　multivariate　statistical　analysis　to　the　Raman　phenotypes,　the　cancer　cells　and　non-tumorous　cells　were　accurately　identified.　The　high　sensitivity　enabled　this　method　to　discriminate　subtle　molecular　changes　in　different　cell　types,　and　the　accuracy　reached　81.2%　with　principal　components　analysis　and　linear　discriminant　analysis.　The　technique　provided　a　rapid,　unlabeled,　and　non-destructive　method　for　the　detection　and　identification　of　various　cancer　types.