The frequency-domain analysis is the most important and applicable signal processing technology in fault diagnosis for machinery. In the literature, a number of frequency-domain-based methods have been applied to detect faults in machinery and each method has its special features. Therefore, selecting appropriate method plays a pivotal role in inspecting defects based on vibration signals. Characteristics of fault-caused vibration signals and frequency-domain-based methods are summarized in this paper using representative examples, so as to establish a rule of selecting appropriate signal methods of extracting vibration features for different mechanical equipments. The results can be used to improve the precision and reliability of many kinds of fault diagnoses for key components in different machinery.