Precision agriculture adoption on farms continues to grow globally on farms.  Today, yield monitors have become standard technologies on grain, cotton and sugarcane harvesters.  In recent years, we have seen industry and even academics leveraging the adoption of precision agriculture technologies to conduct field-scale, on-farm research.  Industry has been a primary driver of the increase in on-farm research globally through the development of software to support on-farm research.  A survey conducted by Ohio State University in 2017 asking soybean farmers that have been using precision agriculture technologies and are using yield monitor data and variable-rate technology indicated 84% of the farmers conducting on-farm research, mainly with industry partners.  However, the design and type of on-farm study varied significantly with some using 5x5 m grids across the field to assign treatments, to other using a block design with blocks of 0.5 ha to 1 ha in size divided into 4 equal squares to assign treatments, and finally some using a strip-trial design.  In these type studies, yield monitor data is used to collect the primary response variable.  The challenge in these design is not the type of design and assign treatments but how one needs to consider yield data assignment to a treatment area.  This notion of errors and limitations within data layers such as yield maps is not taken into consideration within conducting on-farm research.  Therefore, the objective of this study was determine the sensitivity of mass flow sensors for grain yield monitors in relation to plot length for providing yield estimates. Seven plot lengths (treatments) were used:  3.8, 7.6, 15.2, 30.5, 61.0, 121.9, and 243.8 m. There was an additional 30.5 m section on each end of the field to “fill-up” and exit from the plot area to ensure continual flow conditions for the mass flow sensor. Treatment lengths represented smaller plots being utilized in practice, acre-sized plots, and the longest treatment reflecting the natural field variability without intentional yield differences created. Intentional yield differences in maize (Zea mays) were created by alternating strips of 0 and 202 kg N/ha that allowed varying flow conditions to monitor changes in yield. At harvest, corn yield was measured by a commercial combine equipped with two yield monitors, a plot combine, and a weigh wagon.  Estimates from each were compared to determine the sensitivity of the yield monitoring method to detect yield variations along each pass.  Observations indicated that the mass flow sensors could not detect yield differences at scales of 3.8 or 7.6 m. The pattern of yield differences was observed at the 15.2 m and 30.5 m lengths; however, the yield estimates from the mass flow sensors on the commercial combine were not accurate when compared to the plot combine. The estimates from the 61.0 m length more closely matched both the yield variation pattern within a pass and actual yield estimates; however, it was not until 121.9 m where estimates began to converge more frequently and accurately between the flow sensor data and plot combine yield. Therefore, mass flow sensor sensitivity and accuracy did not occur until around 121.9 m establishing the minimum size of treatment area to accurately represent yield.