Natural regeneration of trees may provide a cost‐effective way to restore degraded tropical dry forests, but the factors predominantly affecting natural regeneration success are poorly understood. We performed a multifactor analysis of natural regeneration in an Afromontane forest in Northern Ethiopia. Inventory plots of 400 m², each with a nested 9 m² subplot, were systematically established. All trees larger than 5 cm DBH and natural regeneration were identified and recorded in the plots and subplots, respectively. Variables describing climate, topography, soil, vegetation, and indications of local human degradation were collected. Determinants of regeneration presence and abundance were analyzed using Hurdle negative binomial regressions. Sixty‐nine woody species were represented in the regeneration, with an average regeneration density (SE) of 14,700 (51) ha⁻¹. The presence of regeneration was positively associated with canopy cover (p < 0.001), distance to the road network (p = 0.04), precipitation (p = 0.003) and tree species diversity (p = 0.03), and negatively with Northing (p = 0.003). The regeneration abundance was positively associated with soil quality (p = 0.001), canopy cover (p = 0.04), and precipitation (p = 0.01). Including the abundance of dominant species in the tree canopy and regeneration layers in the regression model did not affect the presence of natural regeneration, but it clearly affected regeneration abundance. In such a model, the regeneration abundance was positively associated with Euclea racemosa seedlings abundance and negatively associated with the basal area of Dodonaea angustifolia (p = 0.02) and Rhus natalensis (p = 0.04). In this model, forest degradation (p = 0.01) negatively affected the late‐successional species regeneration but not the regeneration of the drought‐resistant light‐demanding species. The low representation of late‐successional trees species in the natural regeneration calls for active restoration intervention.