An actin filament bundle approximately 2-5 microns in length is present in the sperm of the blue mussel, Mytilus. In unfired sperm this bundle extends from the midpiece through a canal in the center of the nucleus to terminate on the membrane limiting the inside of the cone-shaped acrosomal vacuole. The bundle is composed of 45-65 actin filaments which are hexagonally packed and regularly cross-bridged together to form an actin paracrystal so well ordered that it has six nearly equal faces. Upon induction of the acrosomal reaction, a needle-like process is formed in a few seconds. Within this process is the actin filament bundle which appears unchanged in filament number and packing as determined by optical diffraction methods. Using fluorescein-conjugated phalloidin we were able to establish that the bundle does not change length but instead is projected anteriorly out of the midpiece and nuclear canal like an arrow. Existing mechanisms to explain this extension cannot apply. Specifically, the bundle does not increase in length (no polymerization), does not change its organization (no change in actin twist), does not change filament number (no filament sliding), and cannot move by myosin (wrong polarity). Thus we are forced to look elsewhere for a mechanism and have postulated that at least a component of this movement, or cell elongation, is the interaction of the actin filament bundle with the plasma membrane.