The effective thermal conductivity of two evacuated expanded perlite powders has been measured at temperatures between 295 K and 1073 K. Since conduction via the gas phase is suppressed, thermal transport occurs only via solid conduction and thermal radiation. Due to thermal contact resistances between the powder particles, solid conduction is very small and radiative heat transport dominates, especially at high temperatures. Applying the guarded-hot-plate (GHP) method to optically thick specimens, the true effective thermal conductivity λeff, which is the sum of solid thermal conductivity λs and radiative thermal conductivity λr from the diffusion model, has been measured. After plotting λeff as a function of third power of absolute temperature and calculating the regression line, λs is obtained from the intercept of the line, and the extinction coefficient for thermal radiation E is determined from its slope. The resulting values are λs = (6.4 ± 1.5) ∙ 10−3 W m−1 K−1 and E = (1600 ± 40) m−1 for the first perlite powder and λs = (3.1 ± 1.3) ∙ 10−3 W m−1 K−1 and E = (5700 ± 350) m−1 for the second. With an effective thermal conductivity below 0.01 W m−1 K−1 up to a mean sample temperature of 473 K, the second material is suitable to realize an economic evacuated powder insulation for medium-temperature applications up to approximately 673 K at the hot side. Both materials have also been investigated with the transient-hot-wire (THW) method. This technique has the advantage of shorter measurement time, but underestimates the effective thermal conductivity according to numerical calculations from literature, especially for samples with 1000 m−1≤E≤ 10 000 m−1. This leads to an apparent extinction coefficient Eapp>E from the λeff vs. T3 plot. Using the same procedure as above, the THW measurements deliver λs = (3.9 ± 2.7) ∙ 10−3 W m−1 K−1 for the first perlite powder and λs = (2.2 ± 1.4) ∙ 10−3 W m−1 K−1 for the second, which agrees with the values obtained from the GHP method. The apparent extinction coefficients are Eapp = (2170 ± 110) m−1 and Eapp = (6730 ± 370) m−1, which corresponds to an overestimation by 35.6% and 18.1%, respectively. Both results are in good agreement with the numerical calculations from literature, which have now been verified experimentally for the first time. Because such calculations can in principle be used to correct experimental THW data, it is possible to extend the applicability of the THW method to materials with 1000 m−1≤E≤ 10 000 m−1, i.e. near the limit of radiation diffusion.