Background: Research on pupils' bullying (1991) and violence (1993) motivated the Dutch Ministry of Education, Culture and Science to initiate a national campaign on school safety. The government campaign was undertaken from 1995 to 2000. Aim: To test for differences in secondary pupils' bullying and violence before and after the campaign while controlling for different contextual variables. Samples: In 1991, a representative survey on bullying was conducted in 36 secondary schools with 1,055 pupils from Year 2 and 4 classes (age range 13?16 years). In 1993, a survey on violent behaviour took place in 71 secondary schools with 1,998 pupils from Year 3 and 4 classes (age range 14?16 years). In 2000, a survey on bullying and violence was conducted in 60 secondary schools with 9,948 pupils from Year 1 to 6 classes (age range 12?18 years). Methods: The data from pupils in identical school years were compared with respect to bullying (1991?2000) and violence (1993?2000). The statistical relations were analysed in two stepwise multiple regression analyses. Year of investigation (1991? 2000, 1993?2000) was the respective dependent variable. The independent variables were pupils' bullying or violence scores, sex, school year, contextual lesson, school, and community variables. Results: In both regression analyses, the contextual lesson and school variables discriminated between the measurement years. Also, compared with 1991, the pupils in 2000 scored lower for being a bully and higher for being bullied directly. Compared with 1993, the pupils in 2000 scored lower for being a victim of intentional damage to property or emotional violence, lower for being a perpetrator of disruptive behaviour in school, and higher for being a perpetrator of intentional damage to property. Conclusions: The differences between the contextual variables measured before and after the campaign reflect changes in educational and instructional situations. Independent of these differ so