Even though it is widely believed that cross-modal neuroplasticity in blindness results in enhanced auditory spatial abilities, a growing number of studies also indicate disadvantages of blind persons for specific, more demanding audiospatial tasks. Here, the effects of blindness on horizontal sound localization were compared for simple localization of target sounds presented in isolation and target localization in an acoustically complex (cocktail-party) situation with multiple distractor sounds. While performances of blind and sighted subjects were similar in the cocktail-party setting, a significant disadvantage of blind subjects was found for simple localization of sound sources presented in the center of the frontal space, with the mean absolute error being more than twice that of controls for targets in the median plane. The latter result could be due to the lack of visual calibration of auditory space, which may, in sighted persons, be most effective in the region of maximum visual acuity. On the other hand, the normal performance of blind persons in cocktail-party localization could be explained by the fact that this task, in addition to genuine spatial analysis, critically involved spectrotemporal analysis of the auditory scene to separate different sources for which superior performance is known from previous research.