This talk gives an overview of a validation framework for complex simulation models applied in STINMOD+, a tax and transfer microsimulation model developed at NATSEM. Errors or bugs in simulation models can be common and hard to detect due to complex internal logic, interactions with large and noisy input data, and unintentional human errors. However, even a minor typo may sometimes lead to substantial deviations in the results, misleading the users. This talk describes the conceptual framework of validation deployed in STINMOD+ development and its implementations in validating the parameters, policies, computation logic, integrations and the overall design of the STINMOD+. It also discusses how the framework can be applied automatically so that the model is continuously being checked for possible errors with limited human interventions. The talk also shares some tips on how the modelling project can be designed with validation in mind.