Identifying the properties of the first generation of seeds of massive black holes is the key to understanding the merger history and growth of galaxies. Mergers between ~100M seed black holes generate gravitational waves in the 0.1–10 Hz band that lies between the sensitivity bands of existing ground-based detectors and the planned space-based gravitational wave detector, the Laser Interferometer Space Antenna (LISA). However, there are proposals for more advanced detectors that will bridge this gap, including the third generation ground-based Einstein Telescope and the space-based detector DECIGO. In this paper, we demonstrate that such future detectors should be able to detect gravitational waves produced by the coalescence of the first generation of light seed black hole binaries and provide information on the evolution of structure in that era. These observations will be complementary to those that LISA will make of subsequent mergers between more massive black holes. We compute the sensitivity of various future detectors to seed black hole mergers, and use this to explore the number and properties of the events that each detector might see in three years of observation. For this calculation, we make use of galaxy merger trees and two different seed black hole mass distributions in order to construct the astrophysical population of events. We also consider the accuracy with which networks of future ground-based detectors will be able to measure the parameters of seed black hole mergers, in particular the luminosity distance to the source. We show that distance precisions of ~30% are achievable, which should be sufficient for us to say with confidence that the sources are at high redshift.