This paper proposes a minimum Lagrange multiplier (LM) unit root test with two structural breaks, addressing the limitations of existing tests that assume no breaks under the null hypothesis. The authors derive the asymptotic properties of the test and evaluate its performance through simulations. The test is shown to be robust to breaks under both the null and alternative hypotheses, ensuring that rejection of the null implies trend stationarity. The test is applied to Nelson and Plosser's (1982) data, and the results are compared with those from the Lumsdaine and Papell (1997) test. The findings suggest that the proposed test provides a more accurate inference in the presence of structural breaks, avoiding the issue of overrejections that can occur with other tests.This paper proposes a minimum Lagrange multiplier (LM) unit root test with two structural breaks, addressing the limitations of existing tests that assume no breaks under the null hypothesis. The authors derive the asymptotic properties of the test and evaluate its performance through simulations. The test is shown to be robust to breaks under both the null and alternative hypotheses, ensuring that rejection of the null implies trend stationarity. The test is applied to Nelson and Plosser's (1982) data, and the results are compared with those from the Lumsdaine and Papell (1997) test. The findings suggest that the proposed test provides a more accurate inference in the presence of structural breaks, avoiding the issue of overrejections that can occur with other tests.