The access to the high resolution digital terrain models (DEM) generated from the data collected by the Shuttle Radar Topography Mission (SRTM) of NASA is freely available to the public. Consequently it has become a source of topographic information which is of great value to scientists involved in geophysical or geodetic analysis. Despite the efforts of the Consultative Group on International Agricultural Research (CGIAR), to validate and complement the information contained in these DEMs (currently offered as version 4.1), they still need to be checked for their accuracy in certain regions of the planet. In this paper, the vertical accuracy of the SRTM 3" version 4.1 DEM was analyzed in several areas of Peru using two sets of control points: the height of the district capitals (the minor politics units) and the heights of the weather and hydrological stations from the National Meteorology and Hydrology Service (SENAMHI) of Peru. The comparison shows that the height differences are independent of the altitude, latitude and longitude of the evaluated points. They are rather related to the aspect of the terrain and to the way the SRTM data were acquired. It shows that the mean square of the height differences at national level was ±20 m for district capitals and ±25 m for the SENAMHI stations. This is slightly larger than the overall accuracy of the SRTM ±16 m.