Wavelet　analysis　method　is　a　promising　tool　for　analysis　of　gamma-ray　spectra　due　to　its　time-frequency　localization　ability　and　multi-scale　resolution　feature.　This　paper　describes　a　method　to　locate　peaks　in　gamma-ray　spectra　based　on　continuous　wavelet　transform　(CWT)　of　the　spectrum　using　Marr　wavelet.　The　theoretical　basis　of　using　Marr　wavelet　to　detect　Gaussian　peaks　was　explained.　A　scale　range,　1　to　FWHMmax+2,　was　then　suggested　to　perform　CWT.　Several　novel　criteria　were　used　to　discriminate　real　peaks　from　noise.　The　detection　ability　of　this　method　was　verified　with　some　measured　spectra.　The　results　indicated　that　Marr　wavelet　did　well　in　locating　gamma-ray　peaks.　©　(2014)　Trans　Tech　Publications,　Switzerland.