This paper theorizes how and why safety climate can be conceived as both a leading and a lagging indicator of safety events (i.e., accidents, injuries). When safety climate is conceived as a leading indicator, a prospective design is utilized and safety climate data are correlated with accidents/injuries that occur in the future. When safety climate is conceived as a lagging indicator, retrospective designs are used in which safety climate data are correlated with prior accidents/injuries. We examine the research literature to reveal that safety climate has been investigated as both a leading and a lagging indicator, but it is usually only examined as one or the other within a given study and has been examined as a lagging indicator most frequently. Consistent with our theorizing, prospective designs yield stronger relationships than retrospective designs, suggesting that safety climate is a better leading indicator than lagging indicator; however, it is clearly both. Implications for safety climate research and study design are discussed. (C) 2009 Elsevier Ltd. All rights reserved.