The speeds of sound in water measured by Del Grosso and Mader [J. Acoust. Soc. Am.52, 1442–1446 (1972)], Kroebel and Mahrt [Acustica35, 154–164 (1976)], and Fujii and Masui [J. Acoust. Soc. Am.93, 276–282 (1993)] were compared. A fairly good agreement was found. A new fifth-order polynomial describing the dependence of the speed of sound in water on temperature (ITS-90) within the limits 0–95 °C was proposed. The importance of the effect of impurities of water on the systematic errors of the measurement results was pointed out. The accuracy of the relative measurements of speed of sound in liquids was discussed in the context of the proper choice of a standard (“true”) value of the speed in water.