The present work studies the trends in drought in northern Algeria. This region was marked by a severe, wide-ranging and persistent drought due to its extraordinary rainfall deficit. In this study, drought classes are identified using SPI (standardized precipitation index) values. A Markovian approach is adopted to discern the probabilistic behaviour of the time series of the drought. Thus, a transition probability matrix is constructed from drought distribution maps. The trends in changes in drought types and the distribution area are analyzed. The results show that the probability of class severe/extreme drought increases considerably rising from the probability of 0.2650 in 2005 to a stable probability of 0.5756 in 2041.