Multilingual Language Control (MLC; Poarch, 2018) is the dynamic process that allows a multilingual individual to engage with an appropriate target language while disengaging from other known languages. According to the Inhibitory Control Model (ICM; Green, 1998), initially all potential target language translations are activated, and when selecting the correct language all other known languages must be inhibited, with the amount of inhibition being proportional to the dominance of the language. Inhibition is often examined through reaction times (RTs), particularly switch costs, or slower naming in switch than nonswitch trials. Meuter and Allport (1999) were the first to show asymmetrical switch costs, with slower L1 switching in bilinguals with different proficiencies in their L1 and L2, a phenomenon that disappeared when the proficiencies in the two languages become more balanced. Inhibition is also examined through errors in lexical retrieval, or intrusions. In a trilingual speeded picture naming study conducted by de Bruin and colleagues (2023), they reported significantly more intrusions from the third language (L3) than the first language (L1) when naming in the second language (L2) providing support for the hypothesis that MLC involves proportional inhibition of the non-target languages relative to their dominance. The present study investigated MLC in 69 English/French/German trilingual participants using language history measures to obtain language history, a lexical decision task to determine relative proficiency, and a speeded picture naming task. The speeded naming task consisted of two balanced conditions, Switch and NonSwitch, across the three languages. Responses were coded for accuracy, intrusion language, and speech onset. Results of this study partially corroborate the results found by de Bruin and colleagues as in the present study we observed more L3 intrusions in L2 target trials in the Switch condition. However, in L1 and L3 target trials, L2 was the main source of intrusions across conditions. Intrusion patterns further showed that languages with similar AoA/proficiency profiles, as well as languages with opposing AoA/proficiency profiles, competed strongly during selection. RT results showed slower lexical retrieval for later-acquired language, while switch costs were only observed during L1 naming. Overall, the RT and intrusion patterns suggest that multilingual language control is shaped by both inhibition and graded lexical activation, with AoA and proficiency jointly determining which languages become the strongest competitors during target-language selection.