Long term use of Maresha plow, a conventional plow pulled by oxen, has resulted in the formation of a restrictive layer thereby limiting water movement and aeration as well as root zone penetration in the soil profile. Several studies have shown that improved tillage practices can positively affect infiltration and aeration resulting in increased rainwater use efficiency and agricultural production. However, rather limited information is available about the use of the Berken plough as a potential alternative for tillage practices. We studied the impact of improved tillage practices on infiltration, erosion, runoff and crop productivity during the rainy period of 2016 in Robit-Bata watershed located in upper Blue Nile, Ethiopia. The experiments were carried out in Maize fields where four tillage treatments were compared: (i) no-till (NT), no ploughing; (ii) conventional (CT), plots tilled three times using oxen driven Maresha, (iii) deep (DT), manual digging up to 60 cm using a mattock and (iv) Berken tillage (BT), plots tilled three times using an oxen driven Berken plough. Soil physical parameters (e.g. penetration resistance, bulk density) where measured before tillage treatment and after the cropping season. Additionally, crop performance (plant height, yield, residual biomass and root depth) and measurements on infiltration, sediment yield and runoff were collected. Tillage mean depth was significantly higher in DT (60 cm) followed by BT (27.58 cm) and CT (18.13 cm). After crop harvesting season, the measured soil penetration resistance was lower than tolerable limit (2MPa) up to depths of 40 cm for both BT and DT. Infiltration capacity also increased from 115.2 mm hr-1 (NT), and 120 mm hr-1 (CT) to 242.4 mm hr-1 (BT), and 261.6 mm hr-1 (DT) (p<0.01) . The mean runoff depth significantly decreased in the DT (29.46 mm) and BT treatments (33.53 mm) as compared to the CT (71.45 mm) and NT (98.77 mm) (p< 0.05). Similarly higher mean soil loss was recorded for the CT (5.5 t ha-1) and NT (6.6 t ha-1) compared to the DT (2.6 t ha-1) and BT (2.6 t ha-1) plots. Deeper tillage in DT and BT treatments as well as the presence of invisible barriers along the contour in Berken system could be the reason for the observed increase in filtration and the reduction of runoff and soil loss. On the other hand, the root length under DT was (> 50cm) followed by BT (>40cm) and NT and CT (both < 30 cm). Grain yield of maize was significantly lower in the NT (2.6 t ha-1) compared to yields measured in the CT (3.8 t ha-1), DT (3.8 t ha-1) and BT (4.0 t ha-1) treatments (p<0.05). Results show that improved tillage practices such as deep tillage or Berken plough could increase permeability and therefore root penetration and agricultural productivity whilst decreasing erosion and runoff in the Ethiopian Highlands. The adoption of these techniques in the Ethiopian highlands could improve the sustainability of rainfed agriculture and reduce the environmental impacts associated with traditional tillage practices.