A new phase unwrapping algorithm based on fringe frequency analysis in Fourier-transform profilometry (FTP) is presented. Its framework consists of digital weighted filtering and modulation ordering. By the frequency-weighted filtering, the analysis of fringe frequency is converted into the analysis of fringe modulation. And based on ordering this filtered fringe modulation, a more optimized phase unwrapping path can be found. When this new algorithm is applied to FTP, the consequent phase unwrapping path is always in the direction from an area in which the local spatial frequency is closer to the fundamental frequency to one in which it is further. Because of the strong correlation between local spatial frequency and the reliability of phase data, this unwrapping path also leads from higher to lower reliability. The experiment shows that the new phase unwrapping algorithm based on fringe frequency is reasonable and effective. Even in the worst case, the error if any, is limited to local-minimum areas.