In a detailed analysis of suicide reporting in 1991 44% of the articles were considered inappropriate and dangerous for an imitation effect. After the launching of media guidelines accompanied by a national press campaign we found in a second analysis in 1994 that the quality of reporting (as defined by the guidelines) had improved significantly. The percentage of articles with a high imitation risk score had dropped to 27.5. In 2004 a new survey over a time period of 12 months was carried out, focusing on print media in the region of Bern. Altogether 513 articles were analysed in detail and the imitation risk score calculated. The percentage of potentially dangerous articles had increased again to 41.5%. There were large differences between the newspapers in the frequency and quality of suicide reporting. In recent years, free daily papers with high circulation figures have appeared, with a high percentage of sensational articles on suicide. We conclude that in order to influence the quality of media reporting on suicide, a new campaign, followed by the ongoing dialogue with editors, is needed.