Land degradation through soil erosion by water is severe in highlands of Ethiopia. In order to curb this problem, the government has introduced sustainable land management interventions in different parts of the country since 1970 and 1980’s. This condition has occurred in Debre Mewi watershed since 1990. However, the impacts of the interventions on soil moisture, soil loss rate and crop yield has rarely been investigated in most parts of Ethiopia. The main objective of this study was to examine the impacts of the soil and water conservation (SWC) practice on soil moisture, soil loss rate, soil organic carbon, and crop yield by comparing treated with nearby untreated watershed in upper Abbay basin, Ethiopia. Soil samples were collected from both treated and untreated watersheds in the upper, middle and lower (slope positions), under different land use types based on the standard procedures for each objective. Soil moisture and leaf area index (LAI) were measured during vegetative, reproductive and ripening stages of the growing season in 2020. Satellite images acquired in 1990 and 2020 were used to identify land use land cover changes (LULC) of the study area. Accordingly, we used measurements from erosion pins in both watersheds across different land use and slope categories to compare soil erosion collected against RUSLE model result. The collected data were analyzed with standard statistical procedures with respect to SWC practice, slope position, land-use type, and soil depth. The crop yield was estimated through empirical model comparing LAI, soil moisture and Normalized difference vegetation index (NDVI) with actual yield. The results reveal that the mean annual soil loss rate in treated watershed was estimated to be 32.47 ton ha-1 year -1 in 1990 and 27.62 ton ha-1 year -1 in 2020, while that of the adjacent untreated watershed was 51.98 ton ha-1 year -1 in 1990 and 67.09 ton ha-1 year -1 in 2020. In addition we found that the absolute net soil erosion from erosion pins was correlated R2 =0.51, p=0.004 with erosion modeled by RUSLE model. In both watersheds, there are significant land use changes over the three decades (1990-2020). The study showed that from 1990 to 2020, eucalyptus plantations have been highly expanded and croplands have been reduced. Because of this, the lowest average annual soil loss estimated due to eucalyptus expansion in Debre Mewi was achieved as 2.85 ton ha-1 year -1 in 1990 and 6.49 ton ha-1 year -1 in 2020. The loss in Sholit watershed reached 8.61ton ha-1 year -1 in 1990 and 7.38 ton ha-1 year -1 in 2020. On the other hand, the study revealed that soil moisture varies between treated and untreated watershed both in terms of space and time. Higher soil moisture was recorded on the grasslands of the treated watershed at a depth of 15-30cm and lowest on croplands and eucalyptus trees at 0-15cm depth. The LAI, a predictor of crop yield, was higher in the treated watershed, boosting for improved crop production. In addition, a higher correlation was observed between the measured and estimated soil moisture across three stages of crop development. The study also showed that the mean value of soil organic carbon contents in treated watershed was relatively greater than that of untreated watershed. The severity of erosion was also less in treated watershed than the untreated one. Treated plots from Debre Mewi watershed stored more soil and organic carbon (SOC) than untreated plots because the practice of SWC plays an essential role in improving soil fertility by maintaining SOC. The research also revealed that there was a statistically significant difference between the treated watershed's topsoil carbon content (0–15 cm) and that of the untreated watershed at 15–30 cm. . Land use and soil depth had a significant impact on the amount of soil organic carbon. In this xviii regard, soil organic carbon was higher in soil depth (0-15 cm), 25 Mg ha-1 , 14.23 Mg ha-1 , 11.25 Mg ha-1 and 9 .65 Mg ha-1 under untreated watershed in mixed forests, eucalyptus plantation, grazing and cultivated lands. In the treated watershed, 25.7 Mg/ha of mixed forest, 15.8 Mg/ha of grazing land, 11.2 Mg/ha of cultivated land, and 10.14 Mg/ha of eucalyptus plantation soil organic carbon were estimated. The result shows that organic carbon in the soil decreases while the depth increases. On the other hand, the results indicated that the reproductive phase is the best time for data measurement to develop a model for pre-harvest crop yield estimation model. The results of modeling suggest that the yield and growth of crops are increased when the soil is moist, and that this sensitivity is greater between sites that have been treated versus untreated. The crops grown on treated sites are generally in better condition, than those grown on untreated sites. The vegetative and ripening phase shows that the lowest correlation between yield and all selected predictors. The average maize yield obtained on the SWC-treated fields (4.67 ton ha-1 ) was high compared to the control (2.82 ton ha-1 ). The yield of millet and teff obtained from the SWC-treated fields (2.47, 2.5) ton ha-1 was also high compared to the control (2.23, 1.82) ton ha-1 . The study recommend that reproductive stage was best time for collecting data for developing crop yield model in Ethiopian highlands. At the local scale, the mean absolute percentage errors (MAPE) of the reproductive stages forecasts ranged from 7% to 16%, 7% to 14%, and 6% to 14% for Maize, Millet and Teff, respectively. This research shows the potential for enhancing the yield prediction model suitable for fields in a variety of climatic conditions by combining NDVI, LAI, and soil moisture. In treated watersheds compared to raw watersheds, the rate of soil loss was generally lower, and more soil moisture and organic carbon were preserved. The overall conclusion was that SWC increased crop production by increasing improving soil moisture. The study's empirical model performed well in agro ecologies similar to those for which it was designed, but results may vary among various crop species and agro ecologies with varying growth calendars. Therefore, it is highly advised to conduct additional testing of the model and indices in other agro ecologies and with different crops. Key words: SWC practice, Treated; Untreated; Leaf area index; Soil moisture