Two distinct patterns of microtubular sliding were observed in rat sperm flagellar axonemes. The particular pattern of sliding was determined by the extraction conditions used to prepare the sperm for axoneme disintegration. Sperm prepared by incubating concentrated suspensions of Triton X-100-extracted sperm at pH 9.0 disintegrated by extruding the doublets and outer dense fibers numbered 4 through 7 in response to Mg-ATP. Sperm prepared by incubating motile Triton X-100-extracted models at 37 degrees C for 1 to 3 hours extruded doublets and outer dense fibers 9, 1 and 2. Axonemes disintegrated by both regimens tended to have doublets 3 and 8 (with their corresponding outer dense fibers), as well as the central pair, in place. In numerous instances, the 3-central-8 complex with outer dense fibers 3 and 8 could be found isolated in midpiece sections prepared from both methods. The 3-central-8 partition was also sometimes seen in isolation in cross-sections of the principal piece where it remained attached to the fibrous sheath. The flagellar remnant produced by extrusion of fibers 4 through 7 under high pH conditions was generally straight or randomly curved. In contrast, the flagellar remnant produced by extrusion of the 9-1-2 bundle of fibers was most often curved into a hook in the midpiece region. While the hook-like configuration was not Ca(2+)-dependent, it may be based on a related mechanism. The sliding of the 9-1-2 group of fibers is a consequence of dynein-tubulin sliding between the 2 and 3 doublets. This sliding pattern appears to be preferentially activated in the motile sperm models in EGTA, but seldom if ever produced sliding in the high-pH-extracted models. We conclude that the 3-central pair-8 complex and associated outer dense fibers form an I-beam-like partition that does not participate in sliding, but acts as a structural foundation for organizing a planar beat. In addition, it is clear that preferential activation of certain dynein arms can be evoked, depending on the treatment regimen employed. This shows definitively that the types of microtubule sliding in the two bend directions are not identical.