Earned Value Management (EVM) is a technique for monitoring, controlling and forecasting the cost and duration of any project. Most research in EVM has focused on anticipating the project final duration and cost. However, few studies have compared the stability of deterministic cost forecasting methods (EAC). Moreover, previous studies have mostly focused on the stability of the cost performance index or CPI. Consequently, the objective of this study is to compare the stability of 15 deterministic cost forecasting methods in EVM. Specifically, we simulate the execution of 4100 artificial projects of different topological structures. Pearson's, Spearman's and Kendall's correlation coefficients are used to measure the stability of EAC methods. The stability of the methods is measured at 10% intervals of project progress. The EAC methods achieve stability from 70% - 80% of project execution onwards. Among the compared methods, the EAC=AC+(BAC-EV) and the EAC=AC+(BACEV)/CPI methods stand out as the more stable. Results of this study will enable project management professionals to take timely corrective measures even at early project execution stages.