Using daily station observations over the period 1951–2013 in a region of south-east Australia, we systematically compare how the horizontal resolution, interpolation method and order of operation in generating gridded data sets affect estimates of annual extreme indices of temperature and precipitation maxima (hottest and wettest days). Three interpolation methods (natural neighbors, cubic spline and angular distance weighting) are used to calculate grids at five different horizontal gridded resolutions ranging from 0.25° to 2.5°. In each case the order of operation in which the grid values of the hottest and wettest day are calculated is varied: either they are estimated from daily grids or from station points and then gridded. We find that the grid resolution-despite showing more regional detail at high resolution – has relatively limited effect when considering regional averages. However, the interpolation method and the order of operation can substantially influence the actual gridded values. And while the difference due to the order of operation is not substantial when using natural neighbor and cubic spline interpolation, it is particularly apparent for indices calculated from daily gridded estimates using the angular distance weighting method. As expected given the high spatial variability of precipitation fields, precipitation extremes are most sensitive to method, but temperature extremes also exhibit substantial differences. For the annual maximum values averaged over the study area, the differences may be up to 2.8 °C for temperature and 60 mm (about a factor 2) for precipitation. Differences are seen most prominently in return period estimates where a 1 in 100 year return value calculated using the angular distance weighting daily gridded method is equivalent to about a 1 in 5 year return value in most of the other methods. Despite substantial differences in the actual values of gridded extremes, analyses suggest that the impact on long-term trends and inter-annual variability is small.