Settlement-based designs for foundations, using subgrade reaction modulus (K s ), is an important technique in geotechnical engineering. Plate load test (PLT) is one of the most commonly applied tests to directly obtain the K s of soils. As the determination of the K s from PLT-especially at depths-is both timeconsuming and costly, it is necessary to develop models that can handle easily measurable characteristics. The suitability of the Group Method of Data Handling (GMDH) polynomial neural network to estimate the K s of clayey soils has been investigated in the present research. In order to develop the GMDH models, 123 data sets from Qazvin, Iran, have been applied. The predictability of the derived equations has been compared with other available equations for clayey soils. The results demonstrated that an improvement in predicting the K s has been achieved. A sensitivity analysis on the best GMDH-based equation shows that the liquid limit (LL) of soils is the most influential parameter on the proposed GMDH model to predict the K s of the clayey soils.