A combination of different methods is described whereby climatological time series can be tested for inhomogeneities using relative homogeneity techniques. The method set includes graphical analysis, a non-parametric Kruskal-Wallis homogeneity test and a Wald-Wolfowitz runs test application to the annual mean difference temperature series between highly correlated stations. A series of Monte Carlo simulation studies was carried out, which determined the inhomogeneity detection efficiencies of these tests. The procedure is statistically rigorous and provides estimates of the time and magnitude of change in the mean. Its application to annual mean temperature differences series for 82 Turkish climate stations indicates that the method set is a valuable tool for testing time series. (C) 1998 Royal Meteorological Society.