Rainfall databases are valuable sources of information for regions with limited observational records. However, the performance of these datasets has typically been assessed at large spatial scales. Therefore, the present study evaluated the performance of three rainfall databases GPCC, MSWEP and TerraClimate for estimating monthly precipitation in the Kardeh watershed. For this purpose, rainfall data from six active ground stations in the studied area were collected and compared with the monthly data of the three databases over the period 1990–2019. To assess their accuracy, four statistical indices including Root Mean of Square Error (RMSE), Nash–Sutcliffe efficiency (NSE), coefficient of determination (R²), and Coefficient of Residual Mass (CRM) were applied. In addition, spider charts were used to examine monthly correlations, box plots to compare the distribution of indices, and a Taylor diagram to analyze the combined error and standard deviation. The results showed that the GPCC database exhibited the best agreement with the observational data, with RMSE, CRM, NSE, and R² values of 7.51, -0.07, 0.71 and 0.74, respectively, demonstrating superior performance compared with the other two datasets. MSWEP provided moderate and acceptable performance, whereas TerraClimate showed lower accuracy in representing the statistical characteristics of regional rainfall. Accordingly, the use of GPCC data is recommended for precipitation studies at the watershed scale in the Kardeh watershed.