In order to illustrate some basic principles of epidemiology and to su
ggest how they can be applied to occupational safety studies, articles
published between 1967 and 1997 concerned with the association of hou
sekeeping and order to safety in the workplace were reviewed. Populati
on studies were identified through electronic databases and manual sea
rches. Five of the studies were descriptive or exploratory, and cited
several factors, including housekeeping, associated with company safet
y. Four studies were quasi-experimental and showed improvements in saf
ety following changes in various behaviors and conditions, including h
ousekeeping. Design weaknesses were found, significantly compromising
the validity of these findings. These included the lack of external co
mparisons, history and selection bias, and failure to control for conf
ounding. Control for other variables was only done partially through t
he design, and no attempt to use multivariate statistical models was m
ade. The comments presented in this review should be seen as a base fo
r guiding future studies in occupational safety. (C) 1998 Elsevier Sci
ence Ltd. All rights reserved.