The constructs from A Neurolinguistic Theory of Bilingualism (Paradis 2004) that have implications for attrition are outlined and predictions are explored: The activation threshold hypothesis predicts that, all other factors being equal, language disuse leads to gradual loss; the most frequently used elements of L2 will replace their (less used) L1 counterparts; comprehension of forms will be retained longer than the ability to produce them. Elements sustained by declarative memory (e.g., vocabulary) are more vulnerable to attrition than those sustained by procedural memory (i.e., phonology, morphosyntax, lexicon). Declarative items are also more susceptible to interference (and hence to attrition by substitution) than implicit items. Pragmatics and conceptual representations are also modified by attrition. Motivation impacts the rate of attrition.