The safety inspectors carried out monitoring visits to 305 building constru
ction sites, and the results were compared with the accident figures of the
same sites. The average number of observations per site was 144, and the o
bserved safety aspects were: working habits, scaffolding and ladders, machi
nes and equipment, protection against falling, lighting and electricity, an
d order and tidiness. Each item was scored as 'correct' if it met the safet
y standards, otherwise the item was scored as 'not correct'. The safety ind
ex was calculated as a percentage of the 'correct' items related to all the
observed items. Only some hours of training were needed for making reliabl
e observations, when the observers already knew the safety standards. Also
the validity proved to be good. The sites were grouped according to the obs
erved safety index in order to limit the huge random variation in the accid
ent rates of single sites. There was a significant correlation between the
observed safety index and the accident rate of the site groups. The sites w
ith the lowest observed safety index had, on average, a three times higher
accident rate than the sites with the highest safety index. The method is u
sed by the site personnel as an internal weekly safety inspection and feedb
ack tool. The state safety inspectors use the method as a means of objectiv
e feedback for the companies. (C) 1999 Elsevier Science Ltd. All rights res
erved.