The mRNAs encoding protamines 1 and 2, transition proteins 1 and 2, and the mitochondrial capsule selenoprotein are translationally repressed with long poly(A) tracts in early spermatids and translationally active with heterogenous shortened poly(A) tracts in late spermatids (Kleene, 1989). In the present study, the spacing of ribosomes on the translationally active forms of each mRNA was calculated from the length of the coding region and the polysome size determined by sucrose gradient and Northern blot analysis. In addition, the rate of initiation of these five mRNAs was compared in the reticulocyte cell-free translation lysate. Our results reveal at least four additional forms of translational control over these mRNAs: (1) The vast majority of the active forms of the transition protein 1 mRNA and both protamine mRNAs sediment with polysomes in which the ribosomes are spaced closer than is typical of mammalian mRNAs (31-38 vs 80-100 bases apart). This implies that the rate of initiation is unusually fast and/or that the rate of elongation is unusually slow. (2) The mitochondrial capsule selenoprotein mRNA also initiates efficiently in vivo and in vitro, but sediments with polysomes in which the ribosomes are spaced wider than on the protamine mRNAs. The small size of these polysomes can be explained by inefficient insertion of selenocysteine residues at UGA codons. (3) The transition protein 2 mRNA is translated on small polysomes and a relatively large fraction sediments as free mRNPs in vivo. In addition, the transition protein 2 mRNA initiates translation inefficiently at high mRNA concentration and efficiently at low mRNA concentration in vitro. These observations suggest that the transition protein 2 mRNA may be translated inefficiently because it is a weak competitor for a limiting initiation factor. (4) Since low levels of cycloheximide fail to increase the polysome loading of transition protein 2 mRNA in culture, active single ribosomes may also be limited in late spermatids.