This paper reports on the application of radar satellite data and Persistent Scatterer Interferometry (PS-InSAR) techniques for the detection of ground deformation in the semi-arid loess region of Lanzhou, northwestern China. Compared with Synthetic Aperture Radar Interferometry (InSAR), PS-InSAR overcomes the problems of temporal and geometric de-correlation and atmospheric heterogeneities by identifying persistent radar targets (PS) in a series of interferograms. The SPINUA algorithm was used to process 40 ENVISAT ASAR images for the study period 2003–2010. The analysis resulted in the identification of over 140000 PS in the greater Lanzhou area covering some 300 km2. The spatial distribution of moving radar targets was checked during a field campaign and highlights the range of ground instability problems that the Lanzhou area faces as urban expansion continues to accelerate. The PS-InSAR application detected ground deformations with rates up to 10 mm a−1; it resulted in the detection of previously unknown unstable slopes and two areas of subsidence.