This article presents the results of a prospective study of occupational safety in the
Swedish construction industry, carried out during the construction of a 1.5 km road tunnel
under central parts of a major Swedish city. The study had a non-experimental design, using
self-reported questionnaire data from four measurement waves, T1–T4, performed with an
interval of seven months from October 2002 to October 2004. The interval was chosen to
counteract recall bias but still allow registration of fluctuations in the measures. To reduce
systematic influence due to seasonal characteristics of the work, we preferred a seven-month
to a six-month interval. The questionnaire was comprehensive, so feasibility, i.e., the number
of measurement waves expected to be acceptable to the respondents with a maintained high
response rate, was also taken into account. The parts of the questionnaire reported on here
covered psychosocial conditions, safety climate, and safety behaviour. Members of the
research team were present during data collection, which took place close to the construction
site, during working hours. The respondents were informed of the purpose and procedures of