The application of 3D visualization technologies to cultural heritage has deeply evolved along years, shifting from the role of an ancillary technical solution to the dignity of a new media. However, the responsibility of building a complete 3D world often relies on a single designer who must be expert in different domains: graphics, interaction, and application domain. In this paper we propose a new contentcentered methodology for the production of 3D interactive environments for cultural heritage, including a more appropriate definition of subjects and skills involved. The methodology has been applied to the design of guided tours presenting cultural events, in order to test its benefits. The experimental results demonstrate that this methodology enables content experts to produce significant 3D experiences collaborating with a team of computer scientists, without concerning about low level implementation details.