Three case-histories are aimed to show possibilities how to use results of dilatometric measurements of relative displacements on rock cracks to assess actual rock slope instability and to make time-predictions of a rock fall. The prediction of a collapse of a sandstone rock wall above an international road in the Northern Bohemia was made 2 months beforehand, and the critical 7-days time-window was met with l-day precision. For an unstable sandstone cliff in the same geographic area, the beginning for the the final phase of a rock fall preparation was foreseen 1 year before. The prognosis of the month entailing immediate rock fall danger was made 5 months beforehand. and then proved using a monitoring method. The time series of a 2 years lasting monitoring enabled the long-time prognosis of rock falling activity from a limestone cliff in Austrian Alps, in 1990. At present time, the prognosis seems to become fulfilled in the next 2 years. These years are making the upper limit of its most probable time window. The needs to improve methods of the medium- and the longtime ranging prognostication to meet the majority of practical demands are briefly discussed.