We demonstrate a method that enables the study of photoluminescence of as-grown nanowires on a native substrate by non-destructively suppressing the contribution of substrate photoluminescence. This is achieved by using polarized photo-excitation and photoluminescence and by making an appropriate choice of incident angle of both excitation beam and photoluminescence collection direction. Using TE-polarized excitation at a wavelength of 488 nm at an incident angle of ∼70° we suppress the InP substrate photoluminescence relative to that of the InP nanowires by about 80 times. Consequently, the photoluminescence originating from the nanowires becomes comparable to and easily distinguishable from the substrate photoluminescence. The measured photoluminescence, which peaks at photon energies of ∼1.35 eV and ∼1.49 eV, corresponds to the InP substrate with zinc-blende crystal structure and to the InP nanowires with wurtzite crystal structure, respectively. The photoluminescence quantum yield of the nanowires was found to be ∼20 times lower than that of the InP substrate. The nanowires, grown vertically in a random ensemble, neither exhibit substantial emission polarization selectivity to the axis of the nanowires nor follow excitation polarization preferences observed previously for a single nanowire.