Cyber-Physical Systems (CPS) are driving the fourth industrial revolution. The more significant impact CPS having on modern world, the more secure CPS must be. But, the complex and heterogeneous nature of CPS makes uncertainty inherent inside CPS. The uncertainty and security challenges of CPS urge for an innovative method for securing CPS. The criticality and complexity of CPS must not allow security to come as an afterthought. Goal: We aim to address these challenges by developing a Model-Based Security Testing approach for CPS under uncertainty. Method: We propose Security-related Uncertainty Model-Based Testing for CPS (SUnMBT4CPS). We build SUnMBT4CPS on a taxonomy of security-related uncertainty in CPS (SUnCPS). Based on the taxonomy, SUnMBT4CPS allows modelling SUnCPS for security testing purposes. From the test models, SUnMBT4CPS generates test cases for exploring the uncertainty boundary of CPS. Results: We have applied SUnMBT4CPS in a case study of smart grids. SUnMBT4CPS allows specifying SUnCPS in the Advanced Metering Infrastructure (AMI) of smart grid. Generated test cases allow discovering security-related uncertainty issues of the AMI. Conclusion: SUnMBT4CPS can support testing the uncertainty of CPS rooted from security issues.