Objective: To assess the effectiveness of hospital-wide patient safety activities in order to improve incident reporting systems. Methods: A questionnaire was administered to all 1039 teaching hospitals in Japan. The number of incident reports was measured by type of profession and the volume of human resources used for patient safety activities during the 6 months in 2006 was inquired. To control the barriers of incident reporting among health care staff and hospital characteristics, we also collected institutional data regarding system design of incident reporting in each hospital, hospital ownership, and hospital size. We measured the amount of patient safety activities for the following domains: meetings and conferences, internal audits, and staff education. The activity volumes were then divided into two categories: above and below the median. Poisson regression with overdispersion was used to perform multivariate analyses of the incident reports. Results: We received 418 replies (40.2%) out of the 1039 institutions. We excluded hospitals with missing data, resulting in final respective samples sizes of 234 and 233 for the number of incident reported by doctor and nurse, respectively. The use of online reporting appeared to increase the number of incident reports by doctors (21.2%, P<0.05), while reducing time required to report the incident (24.3%, P<0.05). Moreover, in hospitals that implemented more staff education for doctors, reporting significantly increased by over 30% (P<0.01). In contrast, nurses were encouraged only by online reporting. Staff education for nurses was not significant factor, though the point estimates tended towards an increased number of incident reports. Conclusion: In accordance with “theory-based” findings in past literature, our findings have shown that patient safety program for staff education might increase the number of incident reports, even if we controlled factors regarding system design of incident reporting, by the analysis of the “empirical” data of a nation-wide multi-centre study.