A multi-sensor system that integrates a combination of GNSS (global navigation satellite system) and other sensors is required to realize the desired robustness of future positioning, navigation and time (PNT) systems. A clear trend in the international development of PNT is modular systems with open and standardized interfaces. To enable such modular systems, the ability to monitor uncertainties and integrity of each module (sensor or subsystem) in the system is required. The goal of this report is to present results from a study of GNSS spoofing suppression with INS (inertial navigation system) support. In the study, several algorithms were implemented and evaluated in the NavToolbox, developed by FOI, which is a software library for multi-sensor fusion in PNT systems. The results show that GNSS spoofing can be detected with the evaluated detection algorithms, supported by INS. Detecting the start of spoofing works well with all evaluated detection methods. Detecting when the spoofing ends, so that the correct GNSS position can be used again, is more challenging especially for the evaluated detectors based on comparison of relative motion between GNSS and INS positions. Innovation-based detection often performs better in the evaluated scenarios than the relative motion-based approach. The innovation-based detector does not use spoofed positions after detection but has a certain delay, due to the length of the observation window, before GNSS is used again after the end of the attack. Spoofed positions that deviate significantly from the true position over a short period of time are relatively easy to detect and thus to exclude from the filter solution. More sophisticated attacks, where the spoofed position deviates gradually over a longer period of time, are more difficult to handle and are only detected some time after the start of the attack. The GNSS position should therefore be excluded from the navigation filter solution even before the attack is detected. How far back the filter should be recalculated depends on several factors, such as the length of the observation window, the nature of the spoofing attack and the quality of the inertial sensors. This is evaluated to some extent in this report, but should be studied further. The evaluation also shows that the filter solution can have a large and rapidly growing error if the filter starts dead reckoning from a point in time that has large uncertainty in speed and orientation, for example when the speed changes. To reduce the risk of initialization errors when GNSS is excluded and only INS is used, the time when the dead reckoning starts should be chosen carefully, for example by identifying previous time intervals when the speed is constant