The estimation of sediments within the reservoir is important during the engineering design and operational phase of the small water impounding system (SWIS). Most of the existing sediment prediction models were developed from a certain number and group of reservoirs (or watersheds) with known attributes, making them of limited use. The present study was conducted to develop localized sediment prediction models that are based on Modified Soil Loss Equation (MUSLE) and suited to the characteristics of SWIS. The data from the recent soil erosion and sedimentation study in six (6) SWIS in Nueva Ecija, namely Villa Isla, Mangandingay, Villa Boado, Maasin, Tibag II, and Alalay Grande SWIS, were utilized in model development. The manual linearity analysis was employed to generate regression coefficient factors (CFs) and models by determining the sites with a greater relationship in terms of measured sediment deposits and MUSLE factors. There were about 19 regression CFs with 15 models, possessing 1-2 independent variables, developed from different land uses and watersheds. The LS factor, having the strongest relationship to soil erosion, was used in tandem with other MUSLE factors to form models with 2 independent variables. The study found an exceptional performance of the developed MUSLE-based models in terms of R2, residual, and mean absolute percentage error (MAPE). The models with 2 independent variables achieved a perfect performance with R2 of 1.0, while no residuals and errors were recorded. The models with 1 independent variable performed well with R2, residual, and MAPE of 0.8036-0.9893, 0.00-2.43, and 7.41-41.05%, respectively.