Ultrasound is widely used in nondestructive testing (NDT). It is the basis of many NDT techniques, such as acoustic emission, phased array and synthetic aperture focusing technique. The attenuation of ultrasound is a key factor that ensures the feasibility of these techniques. Conventional studies of attenuation are mostly conducted in specimens that have dimensions smaller than ultrasound wavelengths. In this type of study, reflections from media boundaries interfere with the wave, resulting in a geometry-dependent attenuation that cannot be applied in another specimen with a different geometry, even if the material is identical. Another factor that impacts on the viability of ultrasound NDT techniques is the spatial uniformity of waves in the concrete sample. To serve as a reference in testing defects, the degree of randomness caused by the material’s heterogeneity should be estimated in conditions free of defects. In this study, a large concrete specimen was fabricated to suppress boundary reflections of the wave, creating an environment less dependent on geometry. Twenty-two PZT (lead zirconate-titanate ceramics) transducers were arranged in a three-by-seven array to transmit and receive ultrasound waves. This experiment sought to measure attenuation over distance and by frequency, as well as analyzing attenuation factors such as geometrical spreading and material absorption. A quadratic trend with good consistency between attenuation and frequency was found, indicating minimal local attenuation. The degree of randomness in the wave was measured by three indices: amplitude, group velocity and the frequency spectrum cross-correlation factor. Frequency dispersion was also described using group velocity. The findings of this study can serve as a reference for other NDT research, disclosing geometry-independent attenuation and the variability of wave characteristics in concrete.