L3 acquisition is influenced by L1-L3 typology, learners' proficiency in those languages, and metalinguistic knowledge. However, the precise patterns of cross-linguistic influence (CLI) in L3 acquisition are still unclear. This study aimed to examine how the abovementioned factors affect learners' sensitivity to subject placement in L3 Italian. We examined L1-Polish speakers (pro-drop language) with L2-English (non-pro-drop) learning Italian as their L3 (pro-drop). Subject placement patterns are shared by Polish and Italian, but differ from English, which may cause positive CLI from L1 or negative CLI from L2. To verify which is the case, 49 Polish undergraduate students of Italian took part in a computerised acceptability judgment task (AJT), which contained exemplars of felicitous and infelicitous subject placement in Italian. We also measured the participants' knowledge of English and Italian, their background languages, and their metalinguistic knowledge. The results showed that the participants lacked sensitivity to the overuse of both overt and null subjects. A subsequent regression analysis revealed L3 proficiency as the only significant predictor of AJT performance. Our data point to some negative CLI from L2 English at lower levels of L3-Italian, or general cognitively-grounded problems with null subjects, which are gradually reduced with the growth of L3 proficiency.