The aim of this research is to study Persian discourse markers in Azerbaijani speakers’ speech. Data was collected from Azerbaijani speakers’ conversations over 18 hours in Ardabil. Investigating data shows that Persian discourse markers, " xob, be hǝ r hal, yǝ ni, vǝ li, belǝ xǝ re, ǝ slǝ n, vaqeǝ n, kolǝ n, vallah, mǝ sǝ lǝ n, xolase, be xatere, ǝ lbǝ te, be qoli, ǝ ma˒ dǝ rhalike˒ bǝ d˒ ҫ ö n˒ dǝ qiqǝ n˒ dǝ rvaqe are used in Azerbaijani conversations. They take place in different positions in a sentence with various discourse functions. The reasons of these substitutions are multifunctionality, lack of equivalents and simplicity of Persian discourse markers. Data analysis based on linguistic and non-linguistic criteria of borrowing/code-switching differentiation shows that the majority of Persian discourse markers have been borrowed by Azerbaijani and some of them appear as code-switched elements in bilinguals’ speech. The findings are in line with Myers-Scotton’ s view describing borrowing/code-switching continuum as a mechanism for accomplishment of borrowing process.