Software requirements management is an essential process to better understand, identify, derive, control and improve system requirements. Typically, requirements are unclear at the beginning and evolve over time. Uncertainties usually produce conflicts among requirements. Rough set analysis (RSA) is a promising technique of granular computing. The emphasis of this paper is on formally defining three software requirements uncertainty problems and on applying RSA to solve these problems. A systematic approach called MATARS was developed for that purpose. We use a modification of a real world software requirements specification (SRS) benchmark example to illustrate main concepts and ideas of the approach.