Discrimination is long seen as a meaningful factor for ethnic inequalities on rental housing markets. Yet empirically, the extent of discrimination is still debatable. For the first time, this article provides a quantitative meta-analysis of field experiments (in person audits and correspondence tests) that were run over the last four decades in the United States, Canada and Europe (N = 71). Special focus is given to a possible inflation of effect sizes by publication bias; to time trends; and to evidence for statistical discrimination. Taken together, nearly all experiments document the occurrence of ethnic discrimination. Effect sizes are inflated by publication bias, but there is still substantial evidence left once the bias is removed. The analysis reveals a consistent decline in the extent of discrimination over time, from moderate levels of discrimination in the 1970s and 1980s, up to only small but still statistically significant levels in the 1990s and 2000s. A significant part of the discriminatory behaviour can be attributed to missing information about the social status of applicants, which supports theories on statistical discrimination. It is discussed how future research could move our knowledge on the underlying mechanisms forward.