Ultrasonic spectroscopy may offer an alternative to imaging methods for the in vivo detection of microscopic cancer. To investigate this potential, a numerical model that incorporates multiple scattering, wave-mode conversion, and hierarchical microstructures was developed to simulate ultrasonic interactions in biological tissue at the microscopic level. Simulated high-frequency (20–75MHz) spectra of up to 2137 cells displayed significant correlations to nucleus diameter and malignant cell infiltration, and indicated as few as 300 malignant cells may be detectable in normal tissue. The results suggest that ultrasonic spectroscopy combined with simulation-based interpretive models may provide real-time histopathology during surgeries, biopsies, and endoscopies.