Clustered robust meta-regression analysis is applied to 109 willingness to pay (WTP) estimates for threatened species from 47 stated-preference studies in 19 countries. Our study updates previous meta-analyses on the topic and tests the effect of important variables not previously considered—species' threat status, use of coloured photographs of species in a survey, and a country's development status, on WTP. We also compared model results obtained from weighting observations by the inverse standard error of WTP and inverse sample size values. Inverse-standard error-weighted model results were more aligned with published research and economic theory and had a better fit than inverse-sample size-weighted model results. Average total present value of WTP was $414/household,1 but variation in reported values was large owing to the survey context. WTP was significantly higher for charismatic and threatened species. Using coloured photographs, or a country's development status did not significantly affect WTP. Average absolute within-sample and out-of-sample transfer errors were estimated to be 17% and 48%, respectively. One-fourth out-of-sample transfers had an error of 10% or less. We discuss limitations and issues in current literature and propose recommendations that will allow future studies to be used in meta-analyses and benefit transfer.