The paper deals with new research findings on pragmatic markers (PMs) use in spoken Russian. The study is based on two speech corpora: “One Day of Speech” (ORD, which contains mainly dialogues), and “Balanced Annotated Collection of Texts” (SAT, which contains only monologues). We explored two annotated subcorpora consisting of 321,504 tokens and 50,128 tokens respectively. The main results are as follows: 1) the extended frequency lists of PMs were formed; 2) PMs, that are frequently used in both types of speech, were identified (e.g., hesitation markers like tam ‘there’, tak ‘that way’), 3) the list of PMs, used primarily in monologue speech, was compiled (in this list there are such PMs as boundary ones znachit ‘well’, nu vot ‘well er’, vs’o ‘that’s all’); 4) the list of PMs, used primarily in dialogues, was made (among such PMs are, for example, “xeno”-markers takoj ‘like’, grit ‘says’ and meta-communicative markers like vidish’ ‘you know’, (ja) ne znaju ‘don’t know’). Particular attention was paid to the variability of pragmatic markers, as well as to complex cases of their identification. Finally, the most common models of pragmatic markers formation (for single-word and multi-word PMs) were revealed.