The　tunable　Q-factor　wavelet　transform　(TQWT)　is　a　discrete　wavelet　transform　realized　by　the　iterative　two-channel　filter　group　and　the　discrete　Fourier　transform　(DFT).　In　this　paper,　the　TQWT　is　applied　to　the　extraction　of　leakage　signals　of　the　water　pipeline　network.　By　using　the　maximum　principle　of　kurtosis,　the　Q-factor　and　the　number　of　decomposition　layer　were　determined　to　overcome　the　disadvantages　of　prefixed　Q-factor　and　number　of　decomposition　layers.　The　combination　of　kurtosis　and　envelope　entropy　was　used　to　select　the　subband　with　abundant　information.　And　then,　correlation　analysis　was　conducted　further　to　locate　the　leakage　point.　Furthermore,　a　laboratory　water　pipe　and　an　outdoor　actual　water　pipe　system　were　built　to　study　the　leakage　signal.　By　analyzing　the　leakage　signal　from　both　experimental　system,　it　can　be　found　that　the　leakage　signal　characteristics　of　different　leakage　points　for　actual　water　pipe　are　all　in　the　first　layer,　and　the　positioning　error　of　each　leakage　point　is　within　1%.　It　indicates　that　TQWT　can　effectively　extract　the　leakage　components　from　the　leakage　signal　of　low　SNR　and　has　a　higher　location　accuracy　of　the　leak　point.