Height estimation for pedestrians is crucial in applications within global navigation satellite system (GNSS)-obstructed or indoor environments, such as firefighting and counter-terrorism, and the fusion of zero velocity update (ZUPT)-based and barometer-based methods (BMs), utilizing motion mode classification, serves as a critical approach. However, current researches still face challenges including insufficient accuracy in scenario-based classification, inadequate adaptation to different motion speeds/modes, and susceptibility to environmental interference. Addressing the aforementioned issues, this article proposes an adaptive zero velocity detection based on the minimum generalized likelihood ratio test statistic per step to enhance the adaptability of ZUPT. In addition, we fuse ZUPT with a barometer to enhance the accuracy of height estimation. To further improve adaptability and mitigate environmental influences when humans are moving, error analyses are conducted on both ZUPT-based and BMs. Through threshold-based and gait analysis approaches, nonscenario-based motion mode classification (no height change, slow height change, and rapid height change) is achieved. Different height estimation strategies [height change constrained to 0 m, BM, and ZUPT-based method (ZM)] are applied to different motion modes to achieve high-performance height estimation. Finally, experimental validation involving multiple individuals, various motion speeds, and scenarios confirms the adaptability of the proposed algorithm.