Deposition of electron cyclotron waves at high-power densities for heating and current drive in a tokamak plasma generates a population of fast electrons and therefore the velocity distribution of the electron population will deviate from a Maxwellian. Apart from the formation of a high energy tail, modifications to the low-energy part of the distribution function have also been observed. This may have consequences for electron temperature measurements by incoherent Thomson scattering, which usually assumes a Maxwellian distribution function to interpret the scattered light spectrum. In order to investigate and quantify such perturbations, a few typical cases of electron cyclotron heating (ECH) and current drive (ECCD) experiments on the Tokamak a Configuration Variable (TCV) have been analysed in detail. The experimental results from the TCV Thomson scattering system have been compared with simulated data assuming different velocity distribution functions. These were obtained either as results of numerical modelling, using the Fokker-Planck code CQL3D, or in the form of a model bi-Maxwellian distribution function based on information from electron cyclotron emission measurements. In some of the investigated cases the deviations from an ideal Maxwellian were significant and systematic errors in the T-e measurements from Thomson scattering could be identified. In particular, for a case with an injected power of 1.35 MW for ECCD, the discrepancies in T, measurements based on different parts of the scattered light spectrum, i.e. different electron velocity classes, were found to be as large as 25-30%. The use of a bi-Maxwellian model distribution function with three free parameters has permitted us to identify the parameter range in which systematic errors exceed the experimental uncertainties of Thomson scattering measurements on TCV.