Various methods for combining station temperature and precipitation time series into regional series are examined. Interpolation of the station series on to regular grid‐boxes of some kind reduces the effects of both spatial and temporal changes in station coverage. Regional time series are best produced by using anomaly or standardized anomaly values rather than the raw values. For temperature, and for spatially coherent regions in terms of precipitation variance, the exact method does not seriously affect the resulting time series, provided anomalies are used, although the magnitudes of trends may differ. For regions with large spatial variations in precipitation variance, the additional step of standardizing the anomaly values is recommended. Both anomaly and standardized anomaly series can be easily transformed back to the original units, although the exact method for doing so can alter the resulting time series in non‐trivial ways.