Learning a second language (L2) implies the incorporation of its words into a lexicon that already contains words of the native language (L1). This chapter considers whether special mechanisms must protect L2 learning in its early stages, how learning the L2 early or late affects the organization of and access to bilingual memory, and how “special” words, like cognates and false friends, are acquired and processed. In addition, it examines the role of competition (inhibition) mechanisms in L2 word learning and L2 word identification according to localist and distributed connectionist models. Simulations with a localist model show that it can account for orthographic aspects of L2 acquisition without assuming any special mechanisms beyond lateral inhibition. The model differentiates the development of the L2 lexicon into stages of sequential or simultaneous L1/L2 learning for various types of words. Simulations with a distributed model show that cognate facilitation and false friend interference effects can be understood not only in terms of an on-line identification perspective but also from a learning perspective. A theoretical comparison of model types concludes the chapter.