The invention discloses a history order cost analysis-based cost prediction method, and belongs to the technical field of manufacturing and informationizing of the metallurgical industry. According to the history order cost analysis-based cost prediction method, the cost is divided according to the cost behaviors and the existing production related data, and a mathematical statistical method is used to estimate the cost developing trend. Proper history cost periods are selected as basic data; if the deadline is too long, the data is relatively old and does not have comparability; and if the deadline is too short, the cost change cannot be reflected so that the prediction is incorrect. Due to the influences of the market prices and prices of the raw/auxiliary materials, no manufacturing costs fluctuate; and by combining the characteristics of the metallurgical industry, a weighted average method is used to analyze the relatively fixed period and change unit cost of each product. During the analysis of the user intention demands, the order receiving weights of key customers and key orders are formulated according to the market conditions, order receiving optimized ranking is carried out on all the orders, and support is provided for the company to decide whether the finance evaluation accepts the intention.