Slope mass movements are the dominant process of sea cliff evolution and are a threat to occupation and human activities located nearby. This fact is compounded with the rapidly increasing human occupation and concentration of economic activities in coastal areas, making cliff hazard assessment a relevant topic. This study was made to assess the effectiveness of a statistically based approach to cliff failure susceptibility assessment in the slowly receding sea cliffs of Mafra County (western coast of Portugal), using the logistic regression method and considering the possible influence of human occupation near the cliffs top, and also to test the true predictive capacity of the susceptibility models produced.The susceptibility assessment was made using number of cliff failure predisposing factors related with geology, geomorphology, wave power and human occupation, which were correlated with a multi-temporal inventory of sea cliff failures for the period 1947–2012, and also with the temporal partition of the cliff failure inventory for the periods 1947 to 1980 and 1980 to 2012, which were used as training and validation sets to test the real predictive capacity of the models generated.The susceptibility models were validated using standard “Receiver Operating Characteristic Curves” (ROC) and the respective “Area Under the Curve” (AUC) which provided encouraging results when compared with the records of past instabilities. However, the predictive capacity computed by the temporal partition of the inventory data was weak, indicating that further studies are required to: 1) assess the monitoring time periods required to provide statistically representative cliff failure inventories in low retreat cliffs, with implication on hazard assessment and long term evolution modeling; 2) cliff failure predisposing factors and their variation along time, namely those related with the human activities and occupation, and also on the cliff failure conditioning and triggering external factors.In spite of this problem, the results obtained with the global inventory data of 1947–2012, which showed good adjustment to the input data, but also on the validation against the time partitions of the inventory, suggest that the selected techniques are viable tools to map sea cliff instability susceptibility and to predict the location of the most critical areas where future instabilities are more likely to occur.