We propose relationship strength in so cial media using folksonomy. We concretely estimate similarity between web pages in social bookmarking services using a tag vocabulary. A social bookmarking service is one of the most famous social media on the Internet. Since the web pages are selected according to users' interests but not their contents, in the services the classification of them is different from one in ordinary content-based approaches. In this pape r we focus relationship st rength estimation among web pages using a tag vocabulary. Avoiding a problem that tags includes some ambiguities among users, a similarity between web pages is defined in each user at first. The similarities are integrated over all users regarding the frequency of evaluation and a variance of the similarities. At last social bookmarking data are represented as a weighted network and it is easy to capture the relationships among all web pages registered in a social bookmarking service. To ev aluate our proposed approach we carry out some experiments using real social bookmarking service, Buzzurl social bookmarking service. Then we confirm the proposed approach is superior to some comparative approaches and our proposed approach can capture more related web pages.