This thesis describes the radio and optical follow-up of 176 sources taken from the 15- GHz 9th Cambridge Survey. The initial motivation for this was to establish the radio source population at high frequency in the expectation that a significant fraction of the sources would prove to be young objects with spectra peaking at high ( rv 5 - 20 GHz) frequencies. Simultaneous radio observations in the GHz to tens of GHz range have provided a continuum radio spectrum for each object and allowed them to be classified on the basis of their simultaneous spectral index. Between one fifth and one quarter of the sources in the flux-limited samples have spectra rising between 1.4 and 4.8 GHz - this is double the fraction of such sources found in surveys conducted at lower frequency. Observations at optical wavelengths are described and the results are analysed. Further radio follow-up at sub-arcsecond resolution has sought to measure the angular size of a sample of all objects, but this has largely failed to resolve the rising-spectrum sources. A 2-3 year study of the variability of the 9C sources at 15 GHz has been carried out and the rising spectrum class of objects is seen to vary more than the flat- and steep-spectrum classes. In addition to the observational work, theoretical models predicting the evolution of infant radio sources are presented and used to make predictions about the observed fractions of rising-spectrum objects in surveys high frequency. The modelling results are found to be consistent with the observati, onal data, all suggest that the risingspectrum class is highly contaminated by objects which are not necessarily young, but instead owe their rising spectra to the effects of relativistic beaming.