The nuclear and conventional power industries have spent enormous resources during the past fifty years investigating the critical heat flux (CHF) phenomenon for a multitude of pool and flow boiling configurations. Experimental CHF data form the basis for the development of correlations and mechanistic models and comparison with them is the sole criterion for a reliable assessment of a correlation or model. However, experimental CHF data are rarely published, remaining in the archives of the authors or in obscure technical reports of an organization. The Purdue University-Boiling and Two-Phase Flow Laboratory (PU-BTPFL) CHF database for water flow in a uniformly heated tube was compiled from the world literature dating back to 1949 and represents the largest CHF database ever assembled with 32,544 data points from over 100 sources. The superiority of this database was proven via a detailed examination of previous databases. A point-by-point assessment of the PU-BTPFL CHF database revealed that 7% of the data were unacceptable mainly because these data were unreliable according to the original authors of the data, unknowingly duplicated, or in violation of an energy balance. Parametric ranges of the 30,398 acceptable CHF data were diameters from 0.25 to 44.7 mm, length-to-diameter ratios from 1.7 to 2484, mass velocities from 10 to 134,000 kg m(-2) s(-1), pressures from 0.7 to 218 bars, inlet subcoolings from 0 to 347 degrees C, inlet qualities from -3.00 to 0.00, outlet subcoolings from 0 to 305 degrees C, outlet qualities from -2.25 to 1.00, and critical heat fluxes from 0.05 x 10(6) to 276 x 10(6) W m(-2). An examination of the parametric distribution of data within the database identified diameters less than 5 mm, mass velocities greater than 10,000 kg m(-2) s(-1), and inlet qualities below -1.0 as those experimental conditions which at present have relatively little CHF data. The combination of these conditions most likely results in subcooled CHF conditions and relatively high CHF values. The PU-BTPFL CHF database is an invaluable tool for the development of CHF correlations and mechanistic models that are superior to existing ones developed with smaller, less comprehensive CHF databases.