The paper investigates the influence of a set of twelve various nature knowledge sharing barriers (shortly, KSBs) on the knowledge sharing speed and quality in the administrative subdivisions of six leading Russian universities. The respondents have been answering the questionnaire including questions regarding the KSBs' perceived strength from the knowledge requester and knowledge holder positions, thus creating four models tested via the IBM SPSS Statistics 22 package as linear regression equations. The testing results are showing statistically significant influence of several KSBs on the knowledge sharing speed and quality. The methodology and results of the study have both practical applicability and theoretical value, as administrative subdivisions are considerably unpopular objects of knowledge sharing research in higher educational institutions, in contrast to the research and teaching subdivisions. Several further research development directions are suggested, considering both the sample extension and changes in the research methodology.