Research reported during the past few decades has revealed the importance for human sound localization of the so-called “monaural spectral cues.” These cues are the result of the direction-dependent filtering of incoming sound waves accomplished by the pinnae. One point of view about how these cues are extracted places great emphasis on the spectrum of the received sound at each ear individually. This leads to the suggestion that an effective way of studying the influence of these cues is to measure the ability of listeners to localize sounds when one of their ears is plugged. Numerous studies have appeared using this monaural localization paradigm. Three experiments are described here which are intended to clarify the results of the previous monaural localization studies and provide new data on how monaural spectral cues might be processed. Virtual sound sources are used in the experiments in order to manipulate and control the stimuli independently at the two ears. Two of the experiments deal with the consequences of the incomplete monauralization that may have contaminated previous work. The results suggest that even very low sound levels in the occluded ear provide access to interaural localization cues. The presence of these cues complicates the interpretation of the results of nominally monaural localization studies. The third experiment concerns the role of prior knowledge of the source spectrum, which is required if monaural cues are to be useful. The results of this last experiment demonstrate that extraction of monaural spectral cues can be severely disrupted by trial-to-trial fluctuations in the source spectrum. The general conclusion of the experiments is that, while monaural spectral cues are important, the monaural localization paradigm may not be the most appropriate way to study their role.