Accurate nuclear data, commonly using evaluated libraries, is essential in many applications, allowing confidence in derived parameters. An approach to assess the confidence with which these data can be used is proposed, not previously reported, comparing nuclear data presented by different evaluations. Variations between evaluations are used as an indication of potential inaccuracies in the nuclear data or evaluation procedure, and the relevant primary literature reviewed more fully. Applying this approach to naturally occurring radionuclides has identified eight radionuclides where the evaluations differ significantly. Where recommended data are supported by a single set of high precision measurements, independent verification of those measurements will increase confidence in the accuracy of the data (214 Bi and 214 Pb). Further measurements should be conducted where the decay schemes are incomplete (228 Ac and 228 Ra). For 40 K, the mean beta energy in all the evaluations has been calculated using an incorrect shape factor, and log ft and branching ratios have been calculated using an inappropriate program. Precise measurements of beta spectra will allow the use of experimentally derived shape factors for the calculation of mean beta energies (40 K and 210 Bi). Parameters used for infinite matrix dose rate and geothermal heat production calculations have been derived for the data discussed here.