In carrier-phase measurements, which are the most precise observations for Global Positioning System (GPS) relative positioning, multipath error is still a factor that interferes with achieving the desired accuracy. Various improvements in receiver and antenna technologies, as well as modeling strategies, have resulted in better ways of coping with this error source. However, errors caused by multipath can be as large as 5 cm, which is not an acceptable accuracy, especially in precise surveying applications like deformation monitoring. In this paper, a full assessment of different wavelets techniques that can be used in multipath mitigation is made to evaluate the optimum way of using wavelets to reduce or remove this type of error. Also, a new approach based on the wavelet detrending technique is introduced to remove carrier-phase multipath error in the measurement domain. To mitigate multipath, GPS double-difference observables are fed to an adaptive wavelet analysis procedure based on high- and low-pass filter decomposition with different levels of resolution. Consequently, the observable sequences are corrected; these corrected observables can then be used to reduce the ambiguity search volume during the initial float solution stage. Meanwhile, double-difference observations with multipath mitigation offer an efficient method for obtaining a better baseline solution.