Coarse--grid global numerical weather simulations or analysis data have to be downscaled, e.g., with nested limited--area models (LAMs), for regional interpretation. Here, the impact of different one--way nesting strategies on precipitation simulations over the European Alps with the LAM ALADIN is studied. The LAM is forced by initial and lateral boundary data derived from ERA40 reanalyses with 120 km horizontal gridspacing and 6 h update interval. The nesting strategies considered include relaxation--based techniques with direct nesting of the high--resolution LAM (horizontal gridspacing Δx = 12 km; domain size 2800 ¿ 2500 km2) or double nesting with an intermediate--resolution nest (Δx = 50 km). Additionally, the impact of a spectral initialization technique is investigated. Results indicate that the considered nesting strategies are comparably successful in terms of precipitation simulation, despite the large resolution jump (120 to 12 km) involved. Thus, the cheapest method in terms of computational resources, i.e., direct nesting, seems to be the most adequate for dynamical downscaling of reanalysis data over complex terrain.