Here we present evidence that strongly suggests that the well-documented phenomenon of A-band shortening in Limulus telson muscle is activation dependent and reflects fragmentation of thick filaments at their ends. Calcium activation of detergent-skinned fiber bundles of Limulus telson muscle results in large decreases in A-band (from 5.1 to 3.3 microns) and thick filament (from 4.1 to 3.3 microns) lengths and the release of filament end fragments. In activated fibers, maintained stretched beyond overlap of thick and thin filaments, these end fragments are translocated to varying depths within the I-bands. Here they are closely associated with fine filamentous structures that also span the gap between A- and I-bands and attach to the distal one-third of the thick filaments. End-fragments are rarely, if ever, present in similarly stretched and skinned, but unstimulated fibers, although fine "gap filaments" persist. Negatively stained thick filaments, separated from skinned, calcium-activated, fiber bundles, allowed to shorten freely, are significantly shorter than those obtained from unstimulated fibers, but are identical to the latter with respect to both the surface helical array of myosin heads and diameters. Many end-fragments are present on grids containing thick filaments from activated fibers; few, if any, on those from unstimulated fibers. SDS-PAGE shows no evidence of proteolysis due to activation and demonstrates the presence of polypeptides with very high molecular weights in the preparations. We suggest that thick filament shortening is a direct result of activation in Limulus telson muscle and that it occurs largely by breakage within a defined distal region of each polar half of the filament. It is possible that at least some of the fine "gap filaments" are composed of a titin-like protein. They may move the activation-produced, fragmented ends of thick filaments to which they attach, into the I-bands by elastic recoil, in highly stretched fibers.