Achieving high workpiece accuracy is a long-term goal of machine tool designers. There are many causes of workpiece inaccuracy, with thermal errors being the most dominant. Indirect compensation (using predictive models) is a promising strategy to reduce thermal errors without increasing machine tool costs. A modeling approach using thermal transfer functions (a dynamic method with a physical basis) has the potential to deal with the issue. The method does not require any intervention into the machine tool structure and its modeling and calculation speed are suitable for real-time applications with results of up to 80% thermal error reduction. Compensation models for machine tool thermal errors based on transfer functions (TFs) were successfully applied on various kinds of single-purpose machines (milling, turning, floor-type, etc.) and implemented directly into various control systems. The aim of this research is to prove the compensation model applicability within the real machining conditions whereas most of the known thermal error models end up with offline verification of their approximation quality. The introduced model of a milling center operates in two machining directions Y and Z and describes thermal errors caused by spindle speed, feed drives, and ambient temperature influences. The model is implemented into a machine tool control system (Fanuc FS31i-B5). The real-time verification upon finishing cutting operation and conditions different from model calibration is discussed in more detail.