For time domain SAR processing algorithms, such as Global Back-projection, interpolation becomes a part of the implementation. In order for the interpolation to yield good results, the oversampling rate used in the sampling of the original data may constitute an important factor. Therefore, the choice of oversampling rate and type of interpolation technique may influence the SAR image quality achieved in images focused with Global Back-projection. The influence of the oversampling rate on the image quality in back-projected SAR impulse response functions is investigated. An attempt to propose useful oversampling rates which yields acceptable SAR image quality is made. Also, nearest neighbor and linear interpolation are compared. Illustration of how the image quality degradation is manifested in the SAR impulse response functions and how image quality measurements considering the azimuth direction exclusively may be used to evaluate the image quality behavior is provided. The investigation is based on an analysis of simulated SAR impulse response functions and the image quality measures of these. The SAR image quality is measured with two common measures, the Integrated Side Lobe Ratio and Peak Side Lobe Ratio. Results are presented for two systems with different fractional and Doppler bandwidths. The results show that the choice of oversampling rate affects the SAR image quality, with significant degradation at the lowest oversampling rates. However, the results indicate that Global Back-projection may offer acceptable SAR image quality already with rather low oversampling rates. Nearest neighbor and linear interpolation display similar SAR image quality. But linear interpolation produces smooth impulse responses, while nearest neighbor interpolation introduces ringing.