In cultured HeLa cells the rates of H3.1 and H3.2 synthesis measured by pulse labeling experiments reflect the steady state content of the two histone variants. This pattern, however, is largely modified when histone translation is carried out in vitro on RNA isolated from the same cell line. In vivo, H3.1 and H3.2 are synthesized approximately at the same rates while the product of H3 mRNA translation in vitro is mostly represented by H3.1 histones. Factors which have so far been invoked for the control of histone messenger RNA stability and translation efficiency are not sufficient to explain our data which in addition indicate that histone H3.1 and H3.2 have different roles in the organization of the genetic material.