Written and sight translation share a comprehension component centered on written input. In sight translation, because of production constraints, the cognitive effort expended in a given span of time is greater than in written translation. Comprehension, transfer, and production processes occur in much shorter periods of time, and this compression of the information processing window has marked consequences for the target production. Translation problems encountered during sight translation are likely to be more disruptive to translation processing. Disruptions will manifest not just as errors or deficiencies in rendering, but as speech disfluencies in the oral performance of the translation. In this paper, we propose that the analysis of speech disfluencies occurring during sight translation performance provides significant information about cognitive phenomena associated with sight translation such as visual interference, as well as about cognitive processes associated with the solution of lexical, syntactic, and strategic translation problems.