Abstract: Micropiles are increasingly used in seismic design or retrofitting for pile foundations. This study aims to investigate the validity of the widely used p-y relationships proposed by the American Petroleum Institute (API) and the ‘m method'; in the Chinese Code for numerical modeling of micropiles in sands. A quasi-static test is performed to reveal the shallow-layer (depth less than 1.0 m) experimental p-y curves for the micropiles in saturated medium dense sands. A piecewise-function based p-y derivation method is adopted to develop the experimental p-y curves. Subsequently, trilinear models for the shallow-layer p-y curves are proposed based on the test results. Further, numerical simulation techniques are used to analyze the validity of the proposed trilinear p-y model, API p-y model and ‘m method'; for predicting the strength and deflection of micropiles in sands. The results show that the API p-y model may slightly underestimate the lateral strength and overestimate the depth-to-maximum bending moment, whereas the proposed shallow-layer trilinear p-y model preferably predicts the global force-displacement relationship, bending moment and lateral deflection of micropiles. In addition, the conventional ‘m method'; is capable to predict the strength and deflection of micropiles. Specifically, a relatively large m value chosen from the recommended range is reasonable for the situation of deflection of small piles at soil surface (less than 6 mm), whereas for the micropiles with a relatively large deflection, the smallest m value among the recommended range is a preferable choice.