ABSTRACT The genome of Semliki Forest virus (SFV) is 11 442 nucleotides with a 5′ cap-structure and a 3′ poly(A) tail of about 100 residues. The genome of the closely related Sindbis virus (SIN) is slightly longer (11 703 nucleotides). The parental RNA is first translated from the 5′ two thirds to yield; nsPl, nsP2, nsP3 and nsP4, which are cleaved from a polyprotein of 2431 amino acids (SFV). The parental genome is copied to a full-length minus strand with poly(U) at the 5′ end. The minus strand is used as template for the synthesis of 42 S RNA in membrane-bound replicative intermediate (RI) structures. In addition to 42 S RNA, a 3′-coterminal subgenomic 26 S mRNA, coding for the structural proteins, is synthesized by internal initiation at the minus strand. Capping and methylation of both plus-strand RNAs occur concomitantly with their synthesis. Analysis of Sindbis virus temperature-sensitive RNA-negative mutants have shown that one complementation group (B) is specifically associated with the synthesis of minus strands. Another, group F, is involved in the polymerization step of both minus- and plus-strand 42 S RNA, and of the 26 S mRNA. The synthesis of minus strands is normally dependent on protein synthesis. There is a shut off of the minus-strand RNA synthesis at about 3h post-infection. This is apparently regulated by a virus-specific protein, represented by the complementation group A. The same protein is involved in the regulation of the initiation of 26 S RNA together with a component represented by group G mutants. Comparative analysis of SFV and SIN RNAs and DI RNAs of both viruses suggests that perhaps only 19 nucleotides from the 3′ end and about 150 nucleotides from the 5’ end are needed for replication of the alphavirus RNAs. In some SIN DI RNAs the proposed secondary structure at the 5’ end is replaced by a cellular tRNAASP suggesting that the secondary structure rather than nucleotide sequence is sufficient for the recognition by the viral polymerase. Even when the primary structure of the four non-structural proteins of both SFV and SIN is known, the correlation of the genetic data with the individual proteins has not yet been possible.