According to generally accepted scanning model proposed by M. Kozak, the secondary structure of 5'-untranslated regions (5'-UTR) of eukaryotic mRNAs can only cause an inhibitory effect on the translation initiation since it would counteract migration of the 40S ribosomal subunit along the mRNA polynucleotide chain. Thus, the existence of efficiently translatable mRNAs with long and highly structured 5'-UTRs is not compatible with the cap-dependent scanning mechanism. It is expected that such mRNAs should use alternative ways of translation initiation to be efficiently translated, first of all the mechanism of the internal ribosome entry mediated by special RNA structures called IRESes (for Internal Ribosome Entry Sites), which have been proposed to reside within their 5'-UTRs. In this paper, it is shown that this point of view is not correct and most probably based on experiments of mRNA translation in rabbit reticulocyte lysate. This cell free system does not reflect correctly the ratio of translation efficiencies of various mRNAs which is observed in the living cell. Using five different mRNAs of similar design which possess either relatively short leaders of cellular mRNAs (beta-globin and beta-actin mRNAs) or long and highly structured 5'-UTRs (c-myc, LINE-1, Apaf-1 mRNAs), we show that the translation activities of all tested 5'-UTRs are comparable, both in transfected cells and in a whole cytoplasmic extract of cultivated cells. This activity is strongly dependent on the presence of the cap at their 5'-ends.