New countermeasures aiming at protecting against power analysis attacks are often proposed proving the security of the scheme given a specific leakage assumption. Besides the classical power models like Hamming weight or Hamming distance, newer schemes also focus on other dynamic power consumption like the one caused by glitches in the combinational circuits. The question arises if with the increasing downscale in process technology and the larger role of static leakage or other harder to model leakages, the pure theoretical proof of a countermeasure’s security is still good practice. As a case study we take a new large ROM-based masking countermeasure recently presented at CT-RSA 2012. We evaluate the security of the scheme both under the leakage assumptions given in the original article and using a more realworld approach utilizing collision attacks. We can demonstrate that while the new construction methods of the schemes provide a higher security given the assumed leakage model, the security gain in practice is only marginal compared to the conventional large ROM scheme. This highlights the needs for a closer collaboration of the different disciplines when proposing new countermeasures to provide better security statements covering both the theoretical reasoning and the practical evaluations.