Building a myofibril from its component proteins requires the interactions of many different proteins in a process whose details are not understood. Several models have been proposed to provide a framework for understanding the increasing data on new myofibrillar proteins and their localizations during muscle development. In this article we discuss four current models that seek to explain how the assembly occurs in vertebrate cross-striated muscles. The models hypothesize: (a) stress fiber-like structures as templates for the assembly of myofibrils, (b) assembly in which the actin filaments and Z-bands form subunits independently from A-band subunits, with the two subsequently joined together to form a myofibril, (c) premyofibrils as precursors of myofibrils, or (d) assembly occurring without any intermediary structures. The premyofibril model, proposed by the authors, is discussed in more detail as it could explain myofibrillogenesis under a variety of different conditions: in ovo, in explants, and in tissue culture studies on cardiac and skeletal muscles.