Three basic models of translation are used in translation research. The first is a comparative model, which aligns translations either with their source texts or with parallel (untranslated) texts and examines correlations between the two. This model is evident in contrastive studies. The second model is a process model, which maps different phases of the translation process over time. This model is represented by communication approaches, and also by some protocol approaches. The third model is a causal one, in which translations are explicitly seen both as caused by antecedent conditions and as causing effects on readers and cultures. The four standard kinds of hypotheses (interpretive, descriptive, explanatory and predictive) are outlined in this chapter and illustrated with reference to the phenomenon of retranslation. Only the causal model can accommodate all four types of hypotheses, and it is hence argued to be the most fruitful model for future development in Translation Studies. Descriptive hypotheses (such as statements about universals or laws) can have explanatory force, but almost all causal influences are filtered through the individual translator’s mind, through particular decisions made by the translator at a given time.