We have previously shown that time-resolved X-ray diffraction studies of the 2-D pattern from isometrically contracting flatfish (turbot) fin muscle have considerable advantages over similar studies of other vertebrate muscles due to the simple lattice and the better long-range order in these muscles (5, 24). Here we show not only that two structurally different myosin head to actin attached states must exist in the crossbridge cycle but that we are also able to define the likely crossbridge configurations in these states. A non-force producing "weak-binding" state is evident from the lead of the (11) equatorial reflections (and actin mass) time-course relative to that of tension (and the (10) equatorial reflection decrease) by about 30 msec. The first myosin layer line at 429 A has a weakened but altered intensity distribution, with no change in axial spacing, in patterns from active muscle. We show this to be consistent with myosin heads binding in the non-specific manner envisaged for the "weak-binding" state. Evidence for the second force-producing attached state, or series of states, comes from the observation of only a small increase in the intensity of the 360 A actin layer line between resting and active muscle patterns. It might be thought that a substantial increase in this layer line would be expected if myosin heads were even transiently attached to the thin filaments in a force-producing state. However, this is not so because internal changes in the structure of the thin filaments in active muscle have the opposite effect of causing this layer line to decrease in intensity. Observation of a small net intensity increase is therefore evidence for myosin head attachment with the symmetry of the actin helix. From the equatorial diffraction pattern, Fourier synthesis maps were computed at 10 msec intervals throughout the isometric tetanus, enabling changes in the mass distribution to be visualised between the force- and non-force producing populations of crossbridges. This difference map shows that in the force-producing state myosin heads have their centres of mass on average at a smaller radius from the thin filament axis compared to the case for non-force producing myosin heads. Since there is good evidence that there is no substantial change in myosin head shape during contraction (30) these observations are consistent with myosin heads swinging on actin as fairly rigid structures.