Eight laboratory mixes and eight field mixes were prepared to study the influence of retempering by adding water to the characteristics of the air-void system, and to study air entrainment when an air-entraining admixture is added significantly after the initial mixing period. For each mix, four sets of samples were cast over a period of 90 min after initial water-cement contact: two before retempering (at 45 min) and two after. The modified point-count method (ASTM C 457) was used to determine the characteristics of the air-void spacing factor, although it often causes a small increase in air content. The results also show that it is possible to entrain an adequate air-void system with a spacing factor of less than 200 mu m if a normal dosage of the air-entraining agent, with a little water, is incorported into the mix 45 min after the initial water-cement contact. An unsatisfactory air-void system can be corrected by adding more of the air-entraining agent after 45 min, but the quantity added must be great enough (i.e., more than 30 to 50 % of the normal dosage) to significanly lower the value of the spacing factor. The results further show that small air-content variations (lower than 2 %) should not be used to predict spacing factor variations.