From a physiological point of view, HMDs presenting an image on each eye are known to offer some advantages comparatively to monocular presentation. Besides the obvious fact that a binocular display provides more `natural' visual perception, it also prevents rivalry and improves several components of the visual function, such as perceptual threshold, contrast sensitivity, and visual acuity. Binocular vision is also a crucial element in depth perception, though its main characteristic, stereopsis, is not yet really used. However, these advantages must be paid by an increased technical complexity and added weight on the head, raising safety related concerns, but also comfort and operational (performance) issues, which imply several tradeoffs. An R&D program funded by the French MOD currently aims to build a night attack HMD for experimental flight tests. Human factor basic requirements were to achieve a head supported mass below 2 kg with minimum encumbrance and to project imagery and symbology on the helmet visor with a large Field of View. The optical and mechanical design was first optimized to allow a head/system resultant CG within the safety limits for ejection. Considering experimental results, a tradeoff is made favoring head mobility rather than seeking stability. Two miniature CRTs are used to display imagery coming either from IR, I2 or TV sources, while symbology is projected monocularly. Consideration of operational needs also implies several tradeoffs at this level.