In bees and other insects, their wings are driven by the alternating actions of two antagonistic flight muscles, dorsal longitudinal muscle (DLM) and dorsoventral muscle (DVM). Here we recorded X-ray diffraction patterns simultaneously from these two muscles during tethered flight, as well as live images of the bees, by using two fast CMOS video cameras at a 5,000/s frame rate. The length changes of the two muscles, as probed by equatorial reflections, are almost perfectly anti-phase, and the DVM is shortest when the wings are upright. In both muscles, force-generating myosin heads, as probed by the 102, 211 and 311 reflection spots, build up slowly during the entire lengthening phase, and their peaks are substantially delayed behind the length. The 111 reflection spot (the first reflection to respond to stretch in skinned flight muscle fiber preparations: 2011 Annual meeting), is conspicuously enhanced in both muscles, well ahead (∼20°) of the aforementioned reflection spots. Previously we interpreted this intensity change to reflect troponin structural changes, but its magnitude and the clear concomitant diminution of the 201 reflection spot are better explained by an azimuthal twist of attached myosin heads. Locating the stretch-sensing mechanism is crucial for understanding the mechanism for stretch activation, and the present results provide evidence that a population of myosin heads respond to stretch in a specific manner. This population is present in the lengthening phase in which fewer heads generate force. Therefore this population is likely to represent low-force myosin heads non-stereospecifically bound to actin. Stretch-induced conversion from low-force to high-force states has been proposed for vertebrate skeletal muscle (Iwamoto, 1995, Biophys. J.), and insects might have evolved to maximally utilize the preexisting function of myosin.