The "Longitudinal Concrete Analysis" dataset contains ultrahigh-speed image sequences of concrete surface cylinders under compression testing in a split Hopkinson pressure bar (SHPB) setup, captured with a Kirana7M camera at frame rates between 1 million and 2 million frames per second (Mfps). A Bézier fitting filtering method has been applied to the raw images to correct for inherent image lag errors of the Kirana7M, enabling proper material analysis based using one-dimensional wave theory. The primary use of this dataset is to estimate the longitudinal wave propagation velocity during the first stress wave transit, achieved through the development of an image-sequence-based approach for selecting stress-wave-induced motion points. Additionally, the dataset is used for the estimation of various material properties, including global deformation, concrete velocity, strain, and stress. Strain measurements obtained through the well-established strain gauge technique—commonly used in split Hopkinson bar testing for material analysis—are also provided within the dataset, for validation of the ultrahigh-speed image sequence analysis.