Reasoning under uncertainty is a key challenge in context aware pervasive systems. In this paper we propose R-CS a situation based context reasoning model that employs ranking technique to rank and order context attributes. Using the proposed ranking technique and available context information, we compute dynamic situation spaces (a collection of contextual attributes that best represent a real world situation) We also propose and incorporate multilevel hierarchical contextual regions into R-CS that enables situation reasoning to be based on one or more dependent context attributes. We present a theoretical approach to compute importance and relevance of newly discovered context attributes which are not defined within the situation space definition by employing the approach of investigating similar neighboring situation spaces. R-CS builds on context spaces theory, a context model based on situation reasoning. We have implemented the proposed algorithms/approaches into R-CS and have validated them by evaluating against context spaces reasoning model.