<p>The beating of the heart is driven by sliding between myosin-containing thick filaments and actin-containing thin filaments and at the single-molecule level by the ‘powerstroke’ in the lever arm of the myosin head, which tilts while its catalytic domain is attached to actin. This tilting lever-arm paradigm was developed before the molecular structure of the thick filament had been determined, and excluded the interactions between the myosin heads and the thick filaments that are now known to stabilise an OFF state of myosin. Here we re-examine the paradigm using measurements of the orientation of two components of the lever arm, the N- and C-lobes of the myosin regulatory light chain (RLC) in heart muscle cells by fluorescence for in situ structure (FISS), comparing them with those in cryo-electron microscopy (cryo-EM) structures of myosin fragments and of the C zone of thick filaments in the OFF state. We show that these FISS and cryo-EM results, combined with those of other structural studies on myosin fragments and muscle cells, can be explained by a modified tilting lever-arm paradigm that includes interactions between the myosin heads and the thick filament during the contractile cycle. In the new model, one head of each myosin dimer remains docked on the surface of the thick filament while its partner ‘working’ head executes the powerstroke and hydrolyses ATP. The interaction between the docked and working heads of the dimer both primes the working head to attach to an appropriate actin monomer in the pre-powerstroke state and re-captures the working head into the dimer after it has hydrolysed ATP.</p>