In this paper we present a method to identify the reflection position in the received Ground-penetrating radar(GPR) signal using short time Fourier transform (STFT) approach. The reflection is identified by using STFT results because at the position of reflection, the STFT result will have a twin-peak phenomenon theoretically and practically. Detailed algorithm analysis, implementation and result are introduced in the paper. The GPR technique provides a continuous, non-destructive tool for thickness profiles. The GPR thickness measurement is based on the principle that the time lag between the direct wave and the reflection from the bottom of the subsurface layer is the function of layer thickness. If low frequency GPR system is used, the penetration depth will increase. On the other hand, the special resolution will be compromised. One of the major problems encountered by the application of GPR system is to locate the starting point of the reflection. However, the direct wave has such high amplitude that it is difficult to identify the reflection, which is a traditional problem in GPR data processing. Mostly one-path signal can be interpreted as the exponentially damped sinusoids signal, and the sub layer reflection signal adds the temporal information into the received signal. Because of the capacity of time-frequency analysis of the STFT, it is a useful tool in GPR signal analysis. Basic STFT is reviewed and algorithms of using STFT are discussed. Examples of the application of STFT are discussed. Examples are obtained from a GPR system designed for thickness measurement of highway pavements. The results show that the STFT process can extract reflection position from GPR waves that mixed with direct wave and can not identified visually.