Resumen
This paper investigated the variability levels of prime cost sums in building projects in order to establish the reliability of such estimates in bills of quantities. It involved a purposive sampling of forty-six executed projects in South-western Nigeria from which data on estimated and final prime cost sums as well as estimated and final contract sums were procured. Ratios interrelating the four variables were generated and their mean values determined. The mean prime cost sum overrun was found to be 41% of the estimated sums. However, a negative relationship between the prime cost sums and the final contract sums was established. Based on observed variations, models were developed to predict any of the ratios in general, and specifically, the practical prime cost range in order to effectively control overall project costs.