||A meta-analysis of 37 studies evaluating the safety effects of public lighting is reported. The 37 studies contain a total of 142 results. The studies included were reported from 1948 to 1989 in 11 different countries. The presence of publication bias was tested by the funnel graph method. It was concluded that there is no evidence of publication bias and that it makes sense to estimate a weighted mean safety effect of public lighting on the basis of the 142 individual results. This is done by the log-odds method of meta-analysis. The validity of the combined results was tested against a number of rival hypotheses. It was concluded that the results are unlikely to have been caused by regression-to-the-mean and secular accident trends. The results were robust with respect to research design, decade of study, country of study, and type of traffic environment studied. The safety effects of public lighting were, however, sensitive to accident severity and type of accident. It was concluded that the best current estimates of the safety effects of public lighting are, in rounded values, a 65 percent reduction in nighttime fatal accidents, a 30 percent reduction in nighttime injury accidents, and a 15 percent reduction in nighttime property-damage-only accidents.