This study proposes that avoidance of a structure by a second language learner can only be claimed for instances where it can be demonstrated that the learner "knows" the avoided language form and that native speakers are expected to use a particular form in that context. Research is reviewed, particularly a study by J. Schachter (1974) in which it is claimed that language differences lead to avoidance. In contrast to this and other studies, the present study claims that the role of contrastive analysis in predicting true avoidance is to show positive transfer potential, or structural similarities, thus contributing to the necessary assumptions about knowledge of the target structure. A third element in a definition of true avoidance is the obligatory environment which in this study refers to semantic or pragmatic features attached to the selection of a particular form contrasted with other paraphrastic options. An experiment in the use of the passive by six native English speakers and six Hebrew-English bilinguals was conducted to test these claims. Subjects were asked to respond to cues containing passive constructions. A comparison of native speaker responses and the bilinguals' responses yielded the following results: (1) the use of the passive by native speakers seems to be topic-dependent, but not that of the bilingual students; and (2) the Hebrew speakers seemed to transfer the lack of usage of the passive in Hebrew to English. It is concluded that true avoidance of a structure in the target language is based, not on ignorance, but on the semantic constraints imposed on that structure in the native language, and that it remains to be seen whether the ability to select a form from among options in a second language is acquireable. (AM)