Spinocerebellar ataxia type 8 (SCA8) is one of a number of dominantly inherited disorders caused by triplet CTG*CAG repeat expansions (2). While investigating the mechanisms of SCA8, Dr. Ranum's lab made the surprising discovery that CAG*CTG expansion constructs express homopolymeric polyglutamine, polyalanine and polyserine expansion proteins without an ATG start codon (3). This repeat associated non-ATG (RAN) translation occurs in transfected cells and lenti-viral transduced cells and brains. Additionally, in vivo mouse and human data demonstrate that RAN-translation across human SCA8 and myotonic dystrophy type 1 (DM1) CAG expansion transcripts results in the accumulation of SCA8 polyalanine and DM1 polyglutamine expansion proteins (3). RAN-translation can occur across CAG expansions in a number of different sequence contexts, but the mechanism of this newly discovered phenomenon, which does not follow the previously described rules of translational regulation, is completely unknown. To gain a better understanding of the mechanisms of RAN-translation and their potential role in microsatellite disorders, I chose to test the hypothesis that RNA sequence variations within and outside of the repeat affect the efficiency of RAN-translation. Data described in my thesis support the following conclusions: 1) the efficiency of RAN-translation in different frames can be positively and negatively affected by the nucleotide sequence within and around the repeat tract; 2) non-ATG translation in rabbit reticulocyte lysates (RRLs) is much less permissive than in HEK293T and N2a cells, initiates with methionine, and requires close cognate start codons (e.g. ATT and ATC); 3) RAN-translation in multiple frames can occur in the presence or absence of an ATG-initiated open reading frame; 4) cellular factors found in HEK293T and N2a cells substantially enhance RAN-translation compared to cell free RRLs; 5) RAN-translation is enhanced across repeat motifs which form hairpin structures. These data have led me to propose two models for RAN-translation: the "stalling model" and the "IRES-like model". The stalling model proposes that scanning ribosomes are stalled by repeat-containing hairpins until translation is initiated and that permissive initiation is increased when hairpins or RNA binding proteins interact with the ribosome and its associated translation initiation factors. The IRES-like model proposes that repeat containing hairpins facilitate RAN-translation by mimicking an internal ribosome entry site (IRES) to recruit the ribosome and initiation factors and initiate translation at non-ATG sites.