An experimental method with headphone virtual reproduction is proposed and a series of experiments to study forward masking effect when the masker and the masked signal are spatially separated in azimuth are conducted. Then, the masking thresholds are compared with those when the masker and the masked signal source are at the same place. The results show that, although both the thresholds of 0° and ±30° sound images increase with the sound pressure level (SPL) of the masker, spatial unmasking may be really observed. The maximum unmasking is as large as 15 dB. This spatial unmasking effect is mainly attributed to better-ear contribution. spatial unmasking, virtual separated sources, forward maskingThe masking threshold decreases when the masked signal is spatially separated from the masker, compared with the situation when they are at the same place. Such a phenomenon is called spatial unmasking and has been studied for many types of stimuli including pure tone [1] , speech [2] , click-trains [3] , and tone complexes [4] . However, the results of these studies are not consistent with each other, and the spatial unmasking ranges from as little as 7 or 8 dB [5] to as much as 24 dB [6] .For broadband noise maskers, spatial unmasking arises primarily from acoustic "better-ear" effects and "binaural" effects. ''Better-ear'' effects lead to unmasking because the signal-to-masker ratio (SMR) generally increases at one ear when the masked signal and the masker are in different directions compared with that when they are in the same direction. Binaural unmasking occurs when the interaural time difference (ITD) and the interaural level difference (ILD) in the masked signal and the masker differ.However, most of these spatial unmasking studies were performed for simultaneous masking and the sounds were replayed with loudspeakers. Forward masking has different characteristics compared with simultaneous masking. For example, with regard to simultaneous masking, at least wideband maskers, the thresholds usually correspond to a constant signal-to-masker ratio; while a given increment in the masker level does not produce an equal increment in the amount of forward