Soil is vulnerable to degradation through deforestation, desertification, salinization, overgrazing and soil erosion. Water erosion causing factors are slope, rainfall, soil type, plant cover, runoff and lack of soil conservation techniques. Present study was carried out in the Dharabi watershed (196 km2) in Chakwal district of Pakistan, where high intensity rainstorms (especially during summer) cause most of the erosion. First part compared the sediment yield from gully and terraced gully land use systems in two pairs of small catchments,-one at Rahna Sadaat and the other at Thoa Bahadar. In each pair, gully system was adjacent to the terraced gully system. Instruments (recording rainguage and water level recorder) were installed to monitor the runoff and sediment yield from all the catchments. Data of soil, land use and vegetation cover were recorded. During three years (2009, 2010, 2011), the sites received 547, 725 and 686 mm rainfall, respectively with corresponding Gumbel probability of exceedance as 0.67, 0.30 and 0.44. In 2010 and 2011, the terraced gully system produced almost 70 per cent lower average sediment yield as compared to gully system. Number of runoff events and runoff coefficient were lower in terrace system. The terraced system produced 64 per cent and 74 per cent lower sediment bound organic carbon losses at two sites with the largest amount of 88.4 kg ha⁻1 from gully catchment. At Rahna Sadaat, average annual loss of sediment bound nitrogen from Terrace-I was 2.5 times lower than from adjacent Gully-I. Loss of available phosphorus was also lower (<1 kg ha⁻1) due to lower content in the soils. Similar results were observed for loss of sediment bound available potassium at this site. Here, terraced catchment decreased annual average loss of extractable K by 3.5 times (0.29 kg xvi ha⁻1 as compared to 1.01 kg ha⁻1 from the gully catchment) during 2010-2011. In the second part, soil erosion plots were established. Soil loss from a cultivated slope on a terrace and an undisturbed slope with natural cover was evaluated to ascertain the impact of cultivation of slopes already converted into terraces. A relatively steep slope (11.2 per cent) and a gentle slope (6.1 per cent) with natural vegetation were compared with a gentle slope (5.8 per cent) on cultivated terrace with existing cropping pattern. Cultivated slope produced the highest (8.96 Mg ha⁻1) soil loss annually as compared to the other undisturbed gentle and steep slopes, viz., 2.08 and 4.66 Mg ha⁻1, respectively. This suggests that a cultivated field on terrace generate more soil loss compared to an undisturbed slope with natural vegetation having similar or higher slope gradient of 11 per cent. It is concluded from the results that the use of a terraced gully system is helpful in reducing losses of sediments and adsorbed nutrients to the reservoir. On the other hand, cultivation on converted sloping terraces increases the soil loss at plot scale. Findings of this study could help making policy decisions regarding the land use change and its downstream impacts.