On the other hand, in the planning phase of works it is not possible to know what will be the actual cost to complete the construction stage, but nevertheless it is in this phase when the practical application of a forecast model makes sense. Although the sample of projects utilized in the development of the models contained only the final costs of construction, the sample of validation projects had both the actual cost and the costs estimated before starting the works. This is because the two samples are derived from totally different databases. But with these data we performed an accuracy comparative analysis of the models, based on the cost type used in the forecasting equation. The aim of this analysis is twofold: (1) we evaluated the statement made by Love et al. (2005), since they considered that cost is an inadequate variable to predict construction time because its final value varies regarding the initial estimate, and (2) taking into account that the regression models have been developed using the actual costs of construction, we check whether its forecasting accuracy is negatively affected when they are tested using the estimated costs of validation projects