Landslide is one of prominent geohazards that frequently affects Indonesia, especially in the rainy season. It destroys not only environment and property, but usually also causes deaths. Landslide monitoring is therefore very crucial and should be continuously done. One of the methods that can have a contribution in studying landslide phenomena is repeated GPS survey method. This paper presents and discusses the operational performances, constraints and results of GPS surveys conducted in a well known landslide prone area in West Java (Indonesia), namely Megamendung, the hilly region close to Bogor. Three GPS surveys involving 8 GPS points have been conducted, namely on April 2002, May 2003 and May 2004, respectively. The estimated landslide displacements in the area are relatively quite large in the level of a few dm to a few m. Displacements up to about 2-3 m were detected in the April 2002 to May 2003 period, and up to about 3-4 dm in the May 2003 to May 2004 period. In both periods, landslides in general show the northwest direction of displacements. Displacements vary both spatially and temporally. This study also suggested that in order to conclude the existence of real and significant displacements of GPS points, the GPS estimated displacements should be subjected to three types of testing namely: the congruency test on spatial displacements, testing on the agreement between the horizontal distance changes with the predicted direction of landslide displacement, and testing on the consistency of displacement directions on two consecutive periods.