The phenomenon of impaired recognition of peripherally presented visual targets, when flanked by similar stimuli, is referred to as crowding. Studies in a two-dimensional space have shown that lateral distances are critical: the extent of crowding depends on eccentricity of the target stimulus and on the spacing between target and flanking stimuli. The question of whether also distances in depth affect crowding was until now usually investigated using virtual depth. However, virtual and real depth differ, for example with respect to the accommodation-vergence alignment and to effects of blur. Thus, we made an attempt to study crowding in real depth. In our experimental setup, real depth is implemented by two screens, observed via a semi-transparent mirror. Thus, moving the two screens along the line of sight allows simultaneous stimulus presentation with real depth differences. In a first validation study with 18 participants, a fixation cross was fixed in a depth of 190 cm. Single and flanked Landolt rings were presented in 2° of eccentricity in the same depth as fixation, or in front of (170 cm), or behind (215 cm) the fixation depth. Results concerning recognition performance show a similar extent of crowding for flanked targets presented in front of, or behind the fixation depth, and flanked targets in the fixation depth. But, concerning reaction time, the difference between isolated and flanked targets was reduced in defocused depths compared to the fixation depth. That is, reaction time toward flanked targets in the fixation depth was higher than in front of, or behind the fixation depth. With the experimental setup, crowding successfully was induced in different real depths. In further studies, the influence of target and flankers in divergent depths on crowding will be investigated.