A specific complex of proteins involved in bacteriophage T4 replication has been visualized by cryoelectron microscopy as distinctive structures in association with DNA. Formation of these structures, which we term "hash-marks" for their characteristic appearance in association with DNA, requires the presence of the T4 polymerase accessory proteins (the products of T4 genes 44, 45 and 62), ATP and appropriate DNA cofactors. ATP hydrolysis by the DNA-stimulated ATPase activity of the accessory proteins is required for visualization of the hash-mark structures. If ATP hydrolysis is stopped by chelation of Mg2+, by dilution with a non-hydrolyzable ATP analogue, or by exhaustion of the ATP supply, the DNA-associated structures disappear within seconds to minutes, indicating that they have a finite and relatively short lifetime. The labile nature of the structures makes their study by more conventional methods of electron microscopy, as well as by most other structural approaches, difficult if not impossible. Addition of T4 gene 32 protein increases the number of hash-mark structures, as well as increasing the rate of ATP hydrolysis. Using plasmid DNA in either a native (supercoiled) or enzymatically modified state, we have shown that nicked or gapped DNA is required as a cofactor for hash-mark formation. Stimulation of the ATPase activity of the accessory proteins has a similar cofactor requirement. These conditions for the formation and visualization of the structures parallel those required for the action of these complexes in promoting the enzymatic activity of the T4 DNA polymerase, as well as the transcription of late T4 genes. Substructure in the hash-marks has been examined by image analysis, which reveals a variation in the projected density of the subunits comprising the structures. The three-dimensional size of the hash-marks, modeled as a solid ellipsoid, is consistent with that of the gene 44/62 protein subcomplex. Density variations suggest an arrangement of subunits, either tetragonal or trigonal, viewed from a variety of angles about the DNA axis. The hash-mark structures often appear in clusters, even in DNA that has a single nick. We interpret this distribution as the result of one-dimensional translocation of the hash-marks along the DNA after their ATP-dependent initial association with, and injection into, the DNA at nicks or gaps.