[en] Differences between power reactor and test reactor data bases have been evaluated. Charpy shift data has been assembled from specimens irradiated in both high-flux test reactors and low-flux power reactors. Preliminary tests for the existence of a bias between test and power reactor data bases indicate a possible bias between the weld data bases. The bias is nonconservative for power predictive purposes, using test reactor data. The lesser shift for test reactor data compared to power reactor data is interpreted primarily in terms of greater point defect recombination for test reactor fluxes compared to power reactor fluxes. The possibility of greater thermal aging effects during lower damage rates is also discussed. 15 refs., 5 figs., 2 tabs