The standard genetic code – the information stored in 64 triplets of nitrogenous bases (codons) in mRNA, which encode 20 proteinogenic amino acids, is nearly universal, with limited exceptions, in all life forms on Earth and it exhibits robustness to harmful effects of mutations in codon sequence. Four hypotheses were put forward in an effort to explain those properties. The frozen accident hypothesis posits that amino acids were coincidentally connected with cognate codons and that the code’s universality is the result of harmful effects of potential alterations. The stereochemical hypothesis explains the connection of amino acids and cognate codons or anticodons by direct stereochemical interactions. The coevolution hypothesis claims that early amino acids were incorporated into the code from abiotic precursors, and the incorporation of late amino acids was concurrent with the evolution of their respective biosynthetic pathways. Error minimization hypothesis proposes that the main selective pressure of code evolution was to obtain greater robustness to effects of mutations, but more recent experiments confirmed that the code’s robustness is most likely a by-product of other selective pressures. Experiments integrating the ideas of the 4 hypotheses with the evolution of the translational apparatus put forward the best possibility of reconstructing the code’s evolutionary path, since none of the hypotheses individually is able to do so. The genetic code today is in a new, in vitro evolutionary stage owing to the development of nonstandard amino acids and synthetic bases. Understanding the code’s past, that is the principles of its evolution, is important in order to accomplish progress in its new evolutionary stage.