Previous assessments of Numerical Weather Prediction (NWP) systems have shown that reanalysis data sets exhibit consistency with the observed climate but tend to be biased, especially at a local scale. In this study, we present a thorough approach to assessing the accuracy of the European Center for Medium‐Range Weather Forecasts, Reanalysis 5 (ERA5) data set compared to low-cost weather stations (Meter ATMOS stations) deployed widely across Kenya as part of the TAHMO weather network.We compare ERA5's performance on temperature (in degrees Celsius) and precipitation (in millimetres) measurements to that of 62 low-cost stations in Kenya for the year 2023. Our assessment performs a detailed comparison of ERA5 reanalysis weather data to the quality-controlled ground truth data from the low-cost weather stations, focusing on error rates and distributions. The detailed data comparison examines the comparative performance of ERA5 and the low-cost stations categorised by distance from the WMO stations within the same climate zones.The Pearson correlation coefficient between the Mean Squared Error (MSE) and the distances is 0.43. An example of the MSE computed for one year on the ERA5 and low-cost station, TA00453, is 23.09 for temperature and 0.44 for precipitation. The moderate correlation between error and distance suggests that as the distance between the weather stations increases, the prediction error also tends to increase, but not in a strong or perfect linear relationship. This could imply that ERA5's accuracy decreases with increased distance from the WMO stations, but other factors could also contribute to the error.Exploring scenarios where low-cost stations complement ERA5 could potentially enhance local weather predictions. Integrating low-cost weather stations with large-scale NWP models like ERA5 can improve the granularity and accuracy of local weather predictions, addressing biases at a local scale and providing more reliable data for various applications.