The main work in bilingual lexicon extraction from comparable corpora is based on the implicit hypothesis that corpora are balanced. However, the different related approaches are relatively insensitive to sizes of each part of the comparable corpus. Within this context, we study the influence of unbalanced comparable corpora on the quality of bilingual terminology extraction through different experiments. Our results show the conditions under which the use of an unbalanced comparable corpus can induce a significant gain in the quality of extracted lexicons.