The aim of this study was to propose the “Within-plot Checks” design in the evaluation of hybrids from initial phases of the Guineagrass breeding program. The design uses common checks in each experimental plot to assess the quality of the microenvironmental in the experiment using plant-based indices associated with row and column arrangements. The design was tested at Embrapa Beef Cattle, Campo Grande, Brazil, where 2,100 hybrids were evaluated in the initial selection phase. Each plot consisted of 32 plants, with two checks and thirty hybrids. Evaluations were carried out on each plant individually and at different times of the year for the traits: Canopy Height; Regrowth Density; and Regrowth Speed. The plant-based index for each plot corresponds to the check mean for each trait. The mixed model methodology was used to test different random and fixed effects. The plant-based index used as a fixed effect was the factor with the greatest influence on the model fitness. Regardless of the trait considered, the association of the plant-based index with the row and column random effects was the one that showed the best performance between the models. Estimating the spatial variance helps to improve the accuracy of the variance components. The narrow and broad sense heritability coefficients were high (> 0.68) for all traits, indicating high prediction accuracy. The “Within-plot Checks” design, uses a plant-based index to characterize plot environmental quality potentially improving prediction accuracy and selection efficiency in the early stages of forage breeding programs.