The problems associated with performing pure-tone threshold measurements in reverberant or diffuse sound fields are illustrated with the use of three-dimensional representations of the sound field within a typical test booth. A microphone mounted on a motorized trolley is used to perform these measurements. A comprehensive comparison is then made of the efficiency with which FM tones, AM tones, damped wave trains, and narrow bands of noise provide a uniform sound field. The conclusion is reached that the bandwidth of the stimulus is the major factor determining the uniformity of the field. A decision about the most appropriate stimulus for sound field work must thus be based on factors other than field uniformity. When the constraints of obtaining suitable spectral distributions, and being able to relate thresholds obtained with complex stimuli to those obtained with pure tones are also considered, FM tones and suitably generated narrow bands of noise appear to be the most suitable stimuli. The selection of suitable parameters is discussed and an Appendix discusses the spectrum and bandwidth of FM signals with different modulation waveforms. The relative accuracy of testing in the direct and reverberant regions in a nonanechoic environment is also discussed.