The Evolutionary Cluster Costing Methodology (ECCM) is proposed for estimating the cost of designing and producing a weapon system at the early design stage. The issue is related to the particular difficulties which system designers often encounter in the absence of cost data on present system details and historically relevant cases associated with the early development phase of a major acquisition project. This is especially relevant in the military environment. In general, the traditional approach for new system cost estimation is to use parametric methods with data from a number of historical cases. However, when there are limited cases it is difficult to establish appropriate and reliable cost models. As an approach to solving this problem, this research has developed ECCM to generate cost characteristics from only a few or even a single existing case to estimate new system costs. The purpose of ECCM is to extract cost characteristics from an existing system by separating it into mutually independent function clusters. Accordingly, ECCM consists of three essentials: function activity cost tables (FACT), an evolutionary clustering methodology, and cost pattern usages. Based on value engineering and system engineering, a system is made up of a particular group of functions, and each function is further supported by certain activities. Because activities can be represented as resources used for supporting related functions, cost employment among functions can be allocated as FACT. As part of the process, a binary incident matrix is constructed, where the values 1 or 0 represent the existence or non-existence of cost interactions between activity and function in FACT. The binary matrix can easily be deduced to represent the most relevant function clusters. To solve the N-P complete combinatorial problem, evolutionary algorithms and proposed cluster evaluation formulae are integrated into the evolutionary clustering methodology. Once the optimal function clusters have been grouped, the costs that interact among functions and activities can be relisted and rated into ratios within each cluster. Cost patterns can then be determined by activity cost ratios from individual clusters. The cost of a new system can be evaluated by considering each similar cluster as a cost parameter because each cluster represents the cost characteristics of a particular function group. Based on the fact that the technology is evolving gradually and the functions in a cluster are related to each other through certain resource relationships, the cost of new systems or products can be estimated by using those clusters. The cost estimates for the new system are obtained through comparing the needs of technologies or values in similar function clusters of existing systems. A case study from three generations of light-sport helicopters has shown that function and assembly clusters can be used to infer the cost of a new design. The results from the case study demonstrate that: 1) the various functions can be clustered to create a certain number of critical purposes, e.g., engine power or structural strength and safety related tasks; 2) ECCM can be used to estimate empirical costs given the absence of detailed design information; 3) the function and assembly similarities of clusters among systems are statistically significant; 4) the differences of cost ratios in related clusters between systems are not statistically significant; 5) the differences between cluster's estimated costs and actual costs in helicopter L-2 or L-3 are not significant statistically; and 6) the differences between assembly's estimated costs and actual costs in L-2 or L-3 are not significant statistically. The cost ratio patterns of individual clusters can be used to target the activity or assembly budgets for developing new systems.