PURPOSES : Roller-compacted concrete pavement (RCCP) is a superstiff-consistency concrete pavement that exhibits excellent strength development owing to a hydration reaction and interlocking aggregates owing to the roller compaction. A zero-slump concrete mixture is generally used. Hence, it is important to control the consistency of the RCCP mixture to prevent the deterioration of the construction quality (such as material separation during paving). The workability of the RCCP is characterized by its consistency and controlled by the Vebe time, whereas a conventional concrete pavement is controlled based on the slump test. The consistency of the RCCP changes over time after concrete mixing owing to delivery, construction time delays, etc. Thus, it is necessary to use the optimum Vebe time to achieve the best construction quality. Therefore, this study aims to develop a Vebe time prediction model for efficiently controlling the consistency of RCCPs according to random time variations. METHODS : A Vebe time prediction model was developed using a multiple linear regression analysis. A dataset of 131 samples was used to develop the model. The collected data consisted of variables with large potential effects on the consistency of the RCCP, such as the water-cement ratio (W/C), sand/aggregate ratio (S/a), water content (ω), water content per unit volume (W), cement (C), fine aggregate (S), coarse aggregate (G), water reducing admixtrue (PNS), air-entraining admixture (AE), delay time (T), air temperature (TEM), and humidity (HUM). In the multiple linear regression analysis, the mentioned parameters were used as the independent variables, and the Vebe time was the dependent variable. The Vebe time prediction models were evaluated by considering the adjusted R2 and p-values. The selection of the model was based on the largest R2 value and an acceptable p-value (p<0.05). RESULTS : The Vebe time prediction model achieved an adjusted R2 value of 64.14% with a significance level (p-value) of less than 0.05. This shows that the predictive model is adequately described for the dependent variable, and that the model is suitable for Vebe time predictions. Moreover, the significance level of the independent variables is less than 0.05, indicating significant effects on the Vebe time (i.e., the dependent variable). CONCLUSIONS : The Vebe time prediction model developed in this study can be used to estimate Vebe times with an R2 of 63.33% between the measured and predicted values. The proposed Vebe time prediction model is expected to be effectively utilized for the quality control of RCCP mixtures. Moreover, it is expected to contribute to achieving good RCCP construction quality.