Recent data on protein crystal growth are reviewed and analyzed with an emphasis to crystal perfection. Since relative supersaturation for protein crystal growth is typically much higher than unity, no considerable detachment of species is expected once joined to the crystal lattice at a kink. Selection of proper molecules and their precise orientation should occur mainly at the stage when they try to join kinks. Sectorial and zonal (striation) structures induced by impurities and other point defects may cause mosaicity even without creation of misfit dislocations. Critical crystal size is estimated al which the dislocation network may be thermodynamically favorable. Since the crystals have isometric shape, this critical size turns out to be about an order of magnitude higher than the critical thickness of a pseudomorphic epitaxial layer. X-ray diffraction rocking curves width of about 0.02 degrees for reflections equivalent to a misfit similar to 3 x 10(-4) should correspond to the critical crystal of ca 0.5 mm, As in epitaxy, if nucleation and spread of dislocations are hindered, the critical size may be essentially bigger or the crystal should crack. At high impurity concentrations, new misoriented blocks may appear by surface nucleation, similar to what should happen in inorganic crystals growing from solutions.