The prediction capability of the 1995 CHF look-up table (Groeneveld D.C., et al., Nucl. Eng. Des. 163 (1996) 1-23) is independently assessed based on the KAIST data base consisting of 10 822 data for uniformly-heated, vertical, round tubes. This confirms the error statistics for the heat balance method reported by Groeneveld et al. and shows overall average and RMS errors of 4.2 and 36.7%, respectively, for the direct substitution method. The new 1995 table shows better prediction capability than the 1986 AECL-UO table (Groeneveld et al., 1986), especially for the low-pressure, low-flow region. The error analysis indicates the length effect even for significantly long tubes. (C) 1997 Elsevier Science S.A.