River blocking caused by landslide dams is a common geological disaster in mountainous regions worldwide, threatening the safety of human lives and infrastructure. However, studies on the formation process of landslide dams and the criteria of river blockage are still at an early stage. This paper built a laboratoryscale experimental apparatus for simulating the formation and evolution process of landslide dams, and then carried out a total of 29 experiments. It analyzed the common evolution stages of river blocking, including landslides, dam formation, and dam breaching in detail. Then, we investigated the influences of the landslide volume, the landslide discharge, and the water flow rate on a criterion and the time length of river blockage. The experimental results suggest that the formation condition of a landslide dam is related to the ratio of the landslide discharge to the water flow rate. Based on this understanding, we proposed a dimensionless River Blockage Criterion (RBC) to judge the formation of landslide dams. The criterion indicates that when RBC > 1.5, a landslide dam may form; otherwise, a landslide dam may not form. Additionally, the time length of river blockage was found to be related to the landslide volume and the water flow rate. Subsequently, we proposed a model for predicting the time length of river blockage and verified it by the 29 experimental results. The Baige River-blocking event, which occurred on the Jinsha River, China, in 2018, was used as a case study to further verify the RBC and the proposed model. The results show that the proposed RBC agreed well with reality, and the percentage difference between the calculated and the actual time length of river blockage was 11%. The findings of this study could contribute to the early warning and emergency response of river-blocking disasters.