Monitoring the health of Midwestern crops throughout the growing season using satellite Normalized Difference Vegetation Index (NDVI) provides valuable information directly related to yield potential. This study’s objective was to quantify the relationship between high-resolution NDVI imagery and high-resolution yield data to predict yield, create high-resolution yield maps, and provide a basis for machine control. Over two years across the Midwest, thirteen test plots were used in this study: three wheat plots, four soybean plots, and six corn plots. PlanetLabs PlanetScope Analytic Ortho Tiles with a spatial resolution of 3.7 m were used to create high-resolution NDVI images that were then corrected for top of atmosphere (TOA) reflectance and masked using plot boundary files. GPS yield data collected in six-meter sub-plots by an Almaco R1 plot combine was overlayed with the generated NDVI image to extract the nearest pixel to the sub-plot’s center latitude and longitude. Simple linear regression (SLR) was used to model the relationship between NDVI and yield. Ten percent of the outliers were removed to account for gauge R&R. The highest coefficients of determination by plot were (0.51, 0.16, and 0.71) for wheat, (0.57, 0.78, 0.49, and 0.29) for soybeans, and (0.44, 0.49, 0.74, 0.49, 0.24, and 0.70) for corn. The model for each crop was the most successful around the period of crop physiological maturity and peak NDVI value. The coefficient of determination as a stand-alone statistical metric is not adequate to evaluate the models’ performance. Little variability in yield or NDVI data will result in a low R2 value, but an exceptional standard deviation of residual error, as well as a reasonable slope, is still possible.