In the next years, the average budget for a professional quality assurance in software projects is predicted to increase up to 28% of the overall budget. In order to control the rising budgets, resources and time for quality assurance activities have to be planned using test effort estimations. In a usual case, an expert performs a test effort estimation using her experience. However, the main drawback of expert-based approaches is a lack of a complete formalization and an empirical evaluation. This PhD Thesis proposes the new method TAQ estimating the effort for software testing activities. The TAQ method contains an algorithmic model consisting of two models: the test complexity model and the test cost driver model. The test complexity model determines, how the test complexity has to be estimated considering the characteristics of each test level. The test cost driver model formalizes the description and the rating of an identified set of influential test cost drivers. These models are combined in a mathematical formula, which can be calibrated on data from previous test projects. The TAQ method is evaluated empirically. The evaluation includes the analysis of completed test projects in the company CRM-IT, an expert questionnaire for test cost drivers, and the calibration of the TAQ method to the CRM-IT context. The formally defined and empirically evaluated TAQ method guarantees the creation of comprehensible and verifiable test effort estimations.