This study presents the use of the method of wavelet transform modulus maxima (WTMM) to detect boundaries of potential field sources. The boundaries of causative sources can be judged by calculating the local modulus maxima of wavelet coefficients at different scales. For the potential field data with noise, the detected boundaries at small scales are easy to be distorted by noise, however, at large scales, the noise can be suppressed greatly and presents more accurate boundary detection results. Therefore, we can get a better boundary judgment by considering the detected boundaries at all scales. Applying the WTMM method to synthetic models and a real data set of Meishan iron deposit, both get a good effect.