Counter-unidirectional anaphora refers to the fact that pronouns precede nouns. This paper analyses and discusses counter-unidirectional anaphora in terms of its syntactic features, generation and interpretation mechanism. Based on a complete description of the data, some relevant theoretical hypotheses are proposed. It is argued that English anaphora differs from Chinese anaphora in terms of syntactic distribution due to their different ways of generation. Nevertheless, they follow the same syntactic-semantic interpretation mechanism. That is, syntactic interpretation follows the Linearity Principle while semantic interpretation the Semantic Orientation Principle; like unidirectional anaphora, counter-unidirectional anaphora of the third person pronoun is subject to the same syntactic and semantic constraints and conforms to the Identifiability Principle, Thematic Hierarchy and Orientation Principle. Hence the so-called counter-unidirectional anaphora is not truly counter-unidirectional, for it strictly abides by the Unilinearity Principle, i.e., in terms of semantic interpretation, nouns precede pronouns, or rather, pronouns are bound in their relevant local domains. Furthermore, counter-unidirectional anaphora is subject to rigid syntactic constraints, i.e., it must conform to the Predicate-Internal Subject Generation Mechanism, and counter-unidirectional anaphora in attributive clauses must also conform to GCR, ECP and island constraints as well as the Binding Principles B and C and satisfy the Prominence Constraint, Feature Compatibility Constraint, and I-within-I Constraint.