Many manufacturers are today striving to offer a large number of value-added PSSs (Product-Service Systems). The increased number of PSS hinders potential buyers from effectively discovering the most suitable PSS to satisfy their personalized requirements. To accurately find the needed or wanted PSS with lower search costs, it is effective to recommend suitable PSS solutions to the right buyers. However, service, a component of PSS, brings more subjective and imprecise information in acquiring users' preferences due to e.g. their different experience and knowledge on services. Moreover, the interactions within user's preferences are often omitted in previous methods, which may lead to inaccurate recommendation results. Therefore, to solve these problems, an innovative method for PSS recommendation is developed. This method explicitly takes into account the environmental aspect of PSSs in question so that a method user can be guided to select an environmentally superior alternative. In addition, rough DEMATEL (Decision-Making and Trial Evaluation Laboratory) is proposed to manipulate the interactions of vague user preferences in multi-criteria weight determination. Furthermore, a rough collaborative filtering approach is developed to make PSS recommendation under vague environment. A case study of elevator PSS recommendation shows the feasibility and potentials of the proposed approach. Theoretically, the new method can produce more reasonable PSS recommendation results by considering the interdependencies between different recommendation criteria. In marketing practice, the method can suggest proposals of new offerings to customers in a proactive manner.