The research described in this paper has addressed itself to the development of on algorithm to forecast potential job costs and throughput times in a job shop. The algorithm utilizes historically repetitive relationships between cumulative job costs find time which are defined as cost curves. The algorithm was developed and evaluated using actual historical data from a large job shop production facility. Evaluation tests showed the cost forecast mechanism to be an effective early predictor of actual total job costs. The throughput time forecast mechanism was found to be less effective as an early predictor, but valuable as an early warning indicator of potential scheduling problems and irregularities.