Ice formation measurements using low temperature calorimetry (CAL) were made on non-air entrained high strength concretes (HSC) before and after exposure to rapid freeze/thaw cycles in water (ASTM C666 proc. A). The purpose was to explore the relationship between water absorption during test, changes in ice formation and deterioration, to investigate the deterioration mechanism of non-air entrained HSC in the ASTM C666 test. CAL results showed zero or very little ice formation in the concretes (w/h = 0.40 and 0.35, 0, 5 and 8 % silica fume) before C666 testing in the temperature range of the test (-20 "C). After the test (DF = 10 -11) significant water absorption had taken place. Only one half or less of the absorbed water was freezeable to -20 "C, i.e. only part of the absorbed water goes into the created cracks and contributes to the deterioration by freezing there. The rest of the water is freezeable at lower temperatures or so tightly bound that it does not freeze at all to -55 "C. The freezeable water to -20 "C after test constitutes 3 -7 vol-% of the cement paste. CAL measurements at intermediate stages of deterioration show very rapid transitions between no and significant amount of freezeable water, and indicate a progressive mechanism of deterioration, starting at the surface in contact with water and moving inward. Apparently very little ice formation can initiate the process and result in major damage. Water storage after freeze/thaw testing results in self healing characterized by significant recovery of dynamic Emodulus and decreased amounts of ice formation. However, compressive strength recovery is much smaller than the recovery of E-modulus.