This theoretical study aims at raising awareness of the existence of lexical false friends (FFs) in English and Arabic as genetically unrelated languages. It also provides a general categorization for FFs from a semantic point of view. A sample of more than fifty FF pairs is examined by contrasting their form, pronunciation and meaning. The analysis reveals that English Arabic FFs are of two types: Chance FFs (occur by coincidence) and semantic FFs(exist via lexical borrowing), the former being more frequent than the latter. This taxonomy is in line with those introduced by Chamizo-Domínguez (2008) and Roca-Varela (2015). Semantic FFs can be total (due to semantic shift, ellipsis, or figurative extension) or partial (through the process of semantic narrowing). Chance FFs are more problematic in the sense that they cause various degrees of embarrassment and confusion due to the fact that some constituents of FF pairs involve taboo and offensive words. Thus, such lexical FFs are considered potential agents that may create misunderstanding, miscommunication and confusion among L2 learners.