A full complement of commercially available fiber property measurements are made on extra long staple (ELS) cottons, starting with the ginned fibers and repeating the measurements at progressive stages in processing leading up to yarn spinning. The behavior of the fiber measurements at different processing stages is observed and the relative performance of these alternative sets of measurements in predicting yarn strength and yarn non-uniformity is examined. This is done for Ne 50 combed ring yarns, Ne 36 carded ring yarns, and Ne 36 rotor yarns. The significant differences in predictive power that are observed cannot be explained by random measurement variations from stage-to-stage. Substantial differences are also seen in cause-and-effect relationships for ring versus rotor spun yarns.