Damage induced by freeze–thaw action in Portland cement concrete pavements can be minimized by ensuring the presence of a well-entrained air-void distribution within its cementitious paste. Several methods exist for assessing the air-void system in fresh concrete, principally ASTM C231-17a, ASTM C173-16, and AASHTO T 395. These methods accurately measure the fresh air content, but this alone is insufficient to assess air-void distribution parameters. ASTM C457-16 enables such an assessment but can only be performed on hardened concrete and consequently does not provide real-time assessment. Hence it was highly desirable to develop a system to accurately predict the air-void distribution properties, principally the spacing factor, of a concrete mixture, in real time, to support quality control/-assurance activities. A set of 271 observations was used to develop statistical models for predicting the spacing factor obtained by the RapidAir-based ASTM C457-16 method. The models were based on paste content and fresh concrete test measurements (predictors), for which least absolute shrinkage and selection operator regression analysis were used to select/reject variable(s). The predicted spacing factors (PSFs) obtained using these models in fresh concrete were evaluated on their ability to differentiate between a well-entrained (i.e., spacing factor <0.2 mm) and coarsely entrained (i.e., spacing factor >0.2 mm) air-void system. The PSFs from the base and final models exhibited 88.5% and 81.1% accuracy, respectively—much higher than that of the SAM number alone (48.4%). A larger fraction of the data set consisted of well-entrained spacing factor observations, which limited validation for coarsely entrained air-void systems.