The contribution that agricultural and accompanying tree species in multistrata agroforestry systems have on biological, physical, and chemical soil properties has been subject to debate.This research evaluated the contribution of trees in coffee based multistrata systems to soil surface protection, soil biota, soil physical properties, runoff and erosion. Part 1 of the study quantified litter thickness, earthworm populations and soil macroporosity in response to land use change, in the Sumberjaya Sub-district, (West Lampung, Indonesia) from in 2001 to 2004. Four Land use systems were compared: (a) remnant forest (control); (b) multistrata shaded coffee with fruit and timber trees as well as nitrogen-fixing shade trees ( Erythrina sububrams and/or Gliricidia sepium ); (c) shaded coffee ( Erythrina sububrams and/or Gliricidia sepium nitrogen-fixing shade trees but less than 5 tree species per plot); and (d) sun coffee ('monoculture') with coffee forming more than 80% of total stem basal area. Plots were selected with an age of 7 - 10 years, in three slope classes: (a) flat (0-100), (b) medium (10-300) and (c) steep (> 300). The mean standing necromass was 6.1, 4.5, 3.8 and 3.0 Mg ha-1 for forest, shade coffee and sun coffee, respectively, without significant influences of slope. Part 2 of the study was a plot-scale erosion experiment, comparing various ages of monoculture coffee systems, various coffee-based systems and natural forest. Plots of 40 m 2 (10 m down slope and 4 m parallel with the contour line) were enclosed by metal sheeting and channeled into a splitter device called "Chin Ong meter". Results show that the mean standing litter stock was 6.1, 4.5, 3.8 and 3.0 Mg ha -1 for forest, shade coffee and sun coffee,respectively, without significant influences of slope. Soil organic carbon contents (Corg) was highest in the forest. The largest annual litter input of 14 Mg ha-1 year -1 was found in the remnant forest, followed by multistrata, shaded and monoculture coffee systems i.e. 9.8, 6.6 and 4.0 Mg ha -1 year -1, respectively. The population density of earthworms in the forest was 50 % lower than that in the multistrata coffee gardens (150 individuals per m 2), but its biomass (31g m -2 ) was twice larger than that in the multistrata coffee gardens. The lowest population density of earthworm was found in the shade coffee system (150 individuals per m2) with a biomass of 7 g m-2. Well-developed coffee-based systems can control soil erosion to about as low as that of forest, but they can not restore surface runoff close to the original forest values. Forest conversion lead to 6 to 10 times increase of the overland flow and to accelerated soil loss particularly in the first two to four years after land clearing. The recovery of a surface litter layer in sun coffee systems can provide protection from erosion with time, but will not be sufficient to restore macroporosity at the level of forest soils, leading to hydrologic alterations that favor overland flow. This research confirmed the role coffee based multi strata systems can play in land rehabilitation.