A feasibility study into the spectral ratio technique of deducing crustal structure at Yellowknife indicates that (a) through the use of long-period teleseismic body waves, the crustal thickness can be determined to within 4 km and the velocity to within 0.6 km/sec of the values obtained by a reversed refraction profile and (b) through the use of short-period body waves, a thin (a few km thick) layer near the surface with a velocity contrast of about 10 per cent with that of the underlying layer can be detected. Theoretical considerations indicate that long-period body waves are not modified, to any significant extent, by absorption in the crust (representative Q values 75-200) but that short-period body waves are. However, estimates of a mean Q for the crust at Yellowknife through the use of short-period body waves are inconclusive, due mainly to inconclusive data—that is, the spectral ratio curves for the different events shown do not agree. The advantages of incorporating multiple phases, such as pP and PcP, into the spectral ratio calculations for long-period body phases are more than offset by the increased intractability in deducing gross crustal features. In the present study, the theoretical composite spectral ratios are restricted to the case where equal amplitudes are assumed for P, pP and PcP, whereas, in the experimental seismograms selected, pP and PcP have smaller amplitudes.