The paper deals with the analysis of request patterns in Swedish and English both in terms of their differences and similarities. The present research is mainly based on P.Brown and S.Levinson’s politeness theory (1978, 1987) as well as the model of social differentiation elaborated by W.Labov, A. Shveizer and V.Zabotkina. The latter is built on two axes of social differentiation of language: stratificational and situational. On the stratificational axis the analysis is carried out taking into account one of the three pragmatic parameters, namely the socio-economic one, where education is one of the crucial criteria for the present paper. The gender and age parameters are outside the scope of this paper. On the situational axis, the role relationships of the interlocutors (symmetric/asymmetric) and the tonality of the situation (official/unofficial) are taken into account. The data for this analysis came from a discourse questionnaire, which consisted of 3 situations requiring requests, and was filled by 50 Swedish and 50 Russian respondents. The following classification has been proposed based on two criteria: 1) the indirectness of the utterances and 2) speaker-hearer -oriented formulas. As a result of the analysis, it becomes obvious that the realization of request patterns in the languages in question occurs mainly with the help of negative politeness strategies modifying their imposition. The speakers tend to use indirect hearer-oriented constructions (questions), semantic minimization, as well as impersonalization. Interrogatives on high-deference level of politeness are characteristic of persons with University degree (UD) in both languages, while statements on low-deference level of politeness are characteristic of representatives without University degree. However, the indirectness of utterances in the languages in question differs. The Swedish language is characterized by the fewer number of politeness strategies per utterance, which lowers its indirectness compared to the English language. The parameter of speaker-heareroriented formulas appeared not to be the leading one in the undertaken research.