In Measurement Based Admission Control (MBAC), the decision of accepting or rejecting a new flow is based on measurements of the current traffic situation. Since MBAC relies on measurements, an in-depth understanding of the measurement error and how it is affected by the underlying traffic is vital for the design of a robust MBAC. In this work, we study how the measurement error impacts the admission decision, in terms of false rejections and false acceptances, and the consequence this has for the MBAC performance. A slack in bandwidth must be added to reduce the probability of false acceptance. When determining the size of this slack, the service provider is confronted with the trade-off between maximizing useful traffic and reducing useless traffic. We show how the system can be provisioned to meet a predefined performance criteria.