The article presents a new method to test for a broken trend in time series data, without needing to know the data's serial correlation. The method is robust and works regardless of whether the data is generated by a specific type of process. Two models are considered: one where only the trend breaks, and another where both the level and trend break. The tests developed in the article are shown to have standard normal distributions and high power to detect trend breaks. In cases where the break date is unknown, the method is still effective. Overall, the tests proposed in the article are more powerful than previous methods in detecting broken trends in data.