Sickle cell anemia results from the formation of hemoglobin S fibers in erythrocytes, and a greater understanding of the structure of these fibers should provide insights into the basis of the disease and aid in the development of effective antisickling agents. Improved reconstructions from electron micrographs of negatively stained single hemoglobin S fibers or embedded fiber bundles reveal that the 14 strands of the fiber are organized into pairs. The strands in each of the seven pairs are half-staggered, and from longitudinal views the polarity of each pair can be determined. The positions of the pairs and their polarities (three in one orientation; four in the opposite orientation) suggest a close relationship with the crystals ofdeoxyhemoglobin S composed of antiparallel pairs of half-staggered strands.The dramatic transformation of erythrocytes caused by the sickle mutation of hemoglobin is due to the association of the hemoglobin S (HbS) molecules into long fibers that align to distort the cell shape. Understanding the self-association of HbS molecules has been the goal of extensive research activity, with studies of the fiber structure by electron microscopy (1) and crystallographic investigations of HbS (2) both providing considerable information. The fiber structure is a complex helical assembly of 14 strands of HbS molecules with a core of 4 strands surrounded by 10 outer strands in a roughly hexagonal packing. The crystals of HbS appear to be related to the fibers, since the valine-,36 residue and other residues implicated in intermolecular contacts of the fibers lie at contacts in the crystals (see refs. 3 and 4). However, the exact relationship between the strands of the fibers and those present in the crystals has remained elusive. Defining this relationship would permit the detailed structural information available for the crystal to be applied to analysis of the complex molecular interactions in the fiber. A study of rare incomplete fibers has demonstrated that the absence of strands occurs in pairs (5), implying the existence of tightly associated double strands. The strands of each pair are staggered by half of a molecular diameter with respect to each other, and hence are possibly related to half-staggered strands also present in the crystal. In the absence of direct information on pairing and strand-pair polarity in the reconstructed images, however, it has not been possible to relate the crystal and fiber structures definitively. A tentative model was proposed that includes the introduction of the transformations necessary to convert linear to helical strands (4).While this model clarified certain aspects of the problem, particularly the role of a-chain mutations that influence fiber formation, additional information on strand pairing and polarity was needed. Such information would serve also to resolve questions raised by other laboratories concerning the possibility of 16-strand structures (6) or a 14-strand structure with other pairing arrangements (7). Improved images of negativ...