Simultaneous interpreting (SI) is a cognitively demanding task that imposes a heavy cognitive load on interpreters. Interpreting into one’s native (A language) or non-native language (B language), known as interpreting directionality, involves different cognitive demands. The cognitive requirements of simultaneous interpreting as well as interpreting directionality affect the interpreting process and product. This current study focused on the lexical features of a specially designed corpus of United Nations Security Council speeches. The corpus included non-interpreted speeches in US English (SubCorpusE), and texts interpreted from Chinese into English (A-into-B interpreted texts, SubCorpusC-E) and from Russian into English (B-into-A interpreted texts, SubCorpusR-E). Ten measures were used to analyze the lexical features of each subcorpus in terms of lexical density, lexical diversity, and lexical sophistication. The three subcorpora were regrouped into two pairs for the two research questions: SubCorpusR-E versus SubCorpusE and SubCorpusR-E versus SubCorpusC-E. The results showed that the interpreted texts in SubCorpusR-E exhibited simpler vocabulary features than the non-interpreted texts in SubCorpusE. In addition, compared with the A-into-B interpreted texts, the B-into-A interpreted texts demonstrated simplified lexical characteristics. The lexical features of the interpreted texts reflect that experienced simultaneous interpreters consciously adopt a simplified vocabulary approach to manage the cognitive load during simultaneous interpreting. This study provides new insights into the cognitive aspects of simultaneous interpreting, the impact of directionality, and the role of lexical strategies. These findings have practical implications for interpreter training, professional growth, and maintaining interpreting quality in diverse settings.