<p>The paper makes an attempt to estimate soil erosion and identify soil erosion susceptibility in Nainital district of India. The soil loss was estimated using Revised Universal Soil Loss Equation (RUSLE). Severe soil loss points extracted from the soil loss map and soil loss controlling factors were integrated to prepare soil erosion susceptibility map using random forest (RF) model. The effectiveness of the model was assessed using performance matrices. The soil loss and soil erosion susceptibility maps were validated using receivers operating characteristic (ROC) curve. The findings revealed that the largest area was under low soil loss class followed by moderate, high and very high classes. Steep slope, high rainfall erosivity, sparse vegetation and inadequate conservation practices have been identified for high and very high soil loss. The soil erosion susceptibility analysis through RF model revealed that nearly 35% area of the district is very highly susceptible due to deforestation, overgrazing and haphazard construction. The discussion with the communities during field work reaffirmed high soil loss and very high soil erosion susceptibility in northern part of the district. Integration of RUSLE and RF model may add a new dimension for devising effective soil conservation measures in spatial information science.</p>