Over the past several years, considerable efforts were devoted in the development of techniques for the detection of steps in climatological series. The main objective was to identify steps caused by non-climatic factors such as instruments relocation and changes in observing procedures and many techniques were based on the use of a reference series to accomplish this. More recently, the techniques have used statistical tests to identify the most probable occurrence of a step and to determine its magnitude and statistical significance. Most of these techniques are quite reliable for the detection of a single step in temperature series. However, very few studies have evaluated the performance of each method for the identification of homogeneous series (series with no steps), and for the detection of multiple steps of different magnitudes. This work presents the comparison of seven methods for their ability in detecting none, one and several steps artificially introduced in simulated temperature time series.