The analysis of landslide processes and consequent damages constitutes an important aspect in risk assessment. The potential reach zones of a landslide can be estimated by analyzing the behavior of past events under particular geological, geomorphological, and climatic conditions. Although landslide risk models have been developed for temperate zones, little information is available for tropical countries, so empirical equations are used without validation. In this study, a dataset comprising characteristic parameters for 123 landslides from the Andean region of Colombia was compiled from the digital inventory of the Colombian Geological Survey Mass Movement Information System (SIMMA). Empirical landslide travel-distance models were developed using simple and multiple regression techniques. The results revealed that the volume of the displaced mass, the slope angle, the maximum landslide height, and geomorphological environment were the predominant factors controlling the landslides travel distances in the study area. Similarly, a strong correlation was found between the planimetric area and landslide volume, validating the model of Iverson et al. (1998) (Iverson et al., in Geol Soc Am Bull 110:972–984, 1998). The proposed models show a reasonable fit between the observed and predicted values, and exhibited higher prediction capacity than other models in the literature. An example of application of the prediction equations developed here illustrates the procedure to delineate landslide hazard zones for different exceedance probabilities.