Noise barriers containing transparent elements may suffer from a reduced shielding effect due to increased multiple reflections between vehicle and barrier. Using mostly simulations, different factors have been identified previously influencing this effect, e.g., size and position of reflecting portions. However, few in situ measurement data have been published. Here, pass-by measurements of a 2 m high, highly absorbing noise barrier are presented for a double-track line during regular operation. The barrier was modified using acoustically reflecting wooden panels placed along the top or the middle of the barrier. Emissions were measured in a close-by reference cross section along the same track without a barrier. Rolling stock comprised different train types. One type of passenger train and cargo trains with homogenous sections of cylindrical tank wagons and flat wagons without load were analyzed. The results show a clear effect of panel positioning, size of the absorbing area, shape of the superstructure, and distance between train and barrier. A simplified single-source model using the 2.5-dimensional boundary element method was shown to model the effect of reflecting panels. However, position and directivity of the source exhibit a considerable influence on the calculated panel effect.