Phase-sensitive　optical　time-domain　reflectometry　(Φ-OTDR)　is　capable　of　detecting　acoustic　emission　induced　small　strain　with　high　sensitivity.　However,　there　is　a　tradeoff　between　sensitivity,　bandwidth　and　detection　range,　which　makes　the　detection　of　a　transient　weak　signal　challenging.　In　this　work,　we　focus　on　transient　weak　signal　detection　using　Φ-OTDR.　To　achieve　this　aim,　we　propose　a　cascaded　statistics-based　signal-processing　framework　in　a　Φ-OTDR　system　to　fetch　the　transient　weak　signal　from　a　noisy　background.　Our　framework　is　based　on　two　key　elements,　including　an　estimator　that　is　derived　based　on　the　probability　characteristic　of　the　Rayleigh　backscattered　light,　and　a　time-frequency　feature　extraction　process　that　maps　the　signal　to　the　time-frequency　domain.　Using　our　statistics-based　signal　processing　Φ-OTDR,　we　demonstrate　experimentally　the　detections　of,　firstly　a　weak　persistent　signal　with　a　magnitude　down　to　4　n　and　a　frequency　up　to　40　kHz,　and　then　a　weak　transient　acoustic　tone-burst　signal.　Our　proposed　scheme　is　promising　for　Φ-OTDR　performance　improvement　particularly　in　weak　signal　detections,　and　it　will　find　new　applications　in　the　future　systems.　©　1983-2012　IEEE.