Knowing the costs of some activity is essential to a financial manager. Within costs it is important to group them into variable costs and fixed costs. The operation of an enterprise generates costs that with the help of the accounting records fail to group costs into the two large categories and a third category of costs arises, namely mixed costs. Mixed costs contain both fixed and variable costs and can only be separated by statistical methods. With the least-squares method, we can make this separation of mixed costs, respecting the conditions imposed by a statistical analysis. Very many use this method without analyzing the parametricity of the data, and the results obtained are of poor quality. In this article we have reviewed the literature on variable, mixed and fixed costs and the statistical model applied. In the research we applied the least-squares regression analysis to the water and sewage operator in Harghita County for the water activity for 2020 and 2019, comparing the results over the two years. The results were also verified with the help of the IBM SPSS analysis program. The conclusion we have reached is that the method of the least-squares is very well applicable for the separation of mixed costs if the data collected at the accounting level are parametric as happens at the regional water operator Harviz S. A., where at the accounting level fixed, mixed and variable production costs are analytically highlighted. The decisions made on the basis of these costs are relevant and enable the undertaking to make the right decisions knowing its break-even point and the fixed costs it can incur. Decisions are also relieved by the fact that fixed costs are highlighted in the two subcategories, namely: short-term fixed costs and long-term fixed costs. As a final conclusion, the decrease in production caused by the reduction in consumption in 2020 did not substantially change the variable costs separated from the mixed costs, so the method used provides support for correct decisions.