Melt-growth theories which are based on the kinetic models of crystal growth are described and compared with the results of computer simulation. Experimental results now available are confronted with the simulation results, and therefrom some doubts are thrown against the original kinetic models for the melt-growth mechanism. Bottle-neck growth is critically reviewed, and is recommended as a method to study the melt-growth mechanism of simple metals. Recent melt-growth theories based on the dislocation model of liquid state are reviewed, and some of the experimental results observed are interpreted properly by the theories.