We conduct a meta-analysis (MA) of around 100 studies valuing nature conservation in Asia and Oceania. Dividing our dataset into two levels of heterogeneity in terms of good characteristics (endangered species vs. nature conservation more generally) and valuation methods, we show that the degree of regularity and conformity with theory and empirical expectations is higher for the more homogenous dataset of contingent valuation of endangered species. For example, we find that willingness to pay (WTP) for preservation of mammals tends to be higher than other species and that WTP for species preservation increases with income. Increasing the degree of heterogeneity in the valuation data, however, preserves much of the regularity, and the explanatory power of some of our models is in the range of other MA studies of goods typically assumed to be more homogenous (such as water quality). Subjecting our best MA models to a simple test forecasting values for out-of-sample observations, shows median (mean) forecasting errors of 24 (46) percent for endangered species and 46 (89) percent for nature conservation more generally, approaching levels that may be acceptable in benefit transfer for policy use. We recommend that the most prudent MA practice is to control for heterogeneity in regressions and sensitivity analysis, rather than to limit datasets by non-transparent criteria to a level of heterogeneity deemed acceptable to the individual analyst. However, the trade-off will always be present and the issue of acceptable level of heterogeneity in MA is far from settled.