To　determine　the　Minbei　Shuixian　tea　from　different　region　quickly　and　non-destructively,　proton　transfer　reaction-time　of　flight-mass　spectrometry　(PTR-TOF-MS)　was　used　to　obtain　chemical　fingerprints　of　the　volatile　compounds　of　113　samples　directly　from　the　surface　of　the　dry　tea　leaves　without　any　sample　pretreatment　in　three　regions　(Wuyishan,　Jianyang　and　Jiannu).　The　raw　MS　data　were　analyzed　by　multivariate　analysis,　including　principal　component　analysis　(PCA)　and　analysis　of　variance　(ANOVA).　In　addition,　soft　independent　modeling　class　analogy　(SIMCA),　K　nearest　neighbor　algorithm　(INN)　and　partial　least　squares　discriminant　analysis　(PLS-DA)　were　used　for　pattern　recognition.　The　experimental　results　demonstrated　that　in　the　positive　ion　mode,　PTR-TOF-MS　combined　with　multivariate　analysis　could　effectively　distinguish　tea　from　different　regions.　Three　principal　components　were　extracted　by　PCA,　and　their　cumulative　contribution　rate　was　84.66%.　The　correct　rates　of　the　three　sets　of　calibration　models　were　89.38%,　100.00%,　100.00%,　and　the　accuracy　rates　of　the　prediction　set　were　83.18%,　96.46%,　95.57%,　respectively.　The　different　region　pattern　recognition　model　were　successfully　established.　It　was　found　that　Shuixian　tea　in　different　regions　of　northern　Fujian　could　be　differentiated　by　PTR-TOF-MS　and　the　results　provided　a　basis　for　establishing　a　fast,　non-destructive　and　sensitive　detection　method　of　tea,　which　could　provide　reference　for　geographical　traceability.