In order to study the nature of amelogenin mRNA, we isolated ameloblast-rich tissue from the unerupted permanent incisor tooth germs of 18-month-old steers and subjected it to guanidine HC1 solubilization for extraction of mRNA. When poly A+ ameloblast RNA was incubated with radioactive deoxynucleotides and reverse transcriptase, four major transcripts were detected with sizes of 1.9, 1.4, 0.7, and 0.4 kb in length. One of the transcripts (0.7 kb) corresponded precisely in length to that predicted from the size of the major in vitro translated amelogenin proteins (27,000 daltons). To determine whether the transcripts did indeed encode amelogenin mRNA, we constructed a λgt11 cDNA library and isolated several amelogenin cDNA's by screening with amelogenin antibody. Four clones were amplified and insert sizes determined by acrylamide gel electrophoresis. Two of the clones had insert sizes of ~ 0.7 kb (λAm 16, XAm 7), and two had insert sizes of ~ 0.4 kb (λAm 11, λAm 4). When the amelogenin cDNA was radiolabeled and used for northern analysis, two species of amelogenin message (0.75 and 0.45 kb) were evident, both of which showed extensive hybridization to λAm 16 (large) and λAm 11 (small) cDNA. These data indicate that: (1) Amelogenin mRNA is heterogeneous in the bovine tooth germ, having two major species 800 and 400 bases long; and (2) the major species of amelogenin share extensive sequence homology. Based on these data, we suggest that at least part of the heterogeneity of amelogenin matrix protein may arise from the production of heterogeneous amelogenin mRNA's that share some common nucleotide sequences.