The need for an empirically defensible means of sequencing instruction appears to have been the primary motivator for research into learning hierarchies. Also, valid learning hierarchies could act as congealing forces in individualizing instructional systems by providing psychological roadmaps for diagnosing students' preinstructional skills and monitoring students' progress through instructional systems. Four methods for generating candidates for learning hierarchies are available: introspection, formal analysis, observation, and statistical "fishing." Experimental transfer of training studies and statistical studies have been used to evaluate posited hierarchies. White recently made significant modifications of these standard evaluation methods to improve the internal validity of research into learning hierarchies. Several external validity issues remain to be resolved before scarce resources should continue to be allocated for additional learning hierarchy research. (Author)