Electromembrane extraction (EME), involving the migration of charged analytes across a supported liquid membrane (SLM) with an external power supply, is a promising sample preparation method in analytical chemistry. However, the presence of boundary double layers at the SLM/solution interfaces often restricts extraction efficiency. To avoid this, the current work proposed an ultrasound-assisted EME (UA-EME) method based on a novel type of supported semi-liquid membrane (SsLM). The characterizations showed that the SsLM was stable under ultrasound conditions. Ultrasound was found to reduce the boundary double layers and thus increase the mass transfer. Major operational parameters in UA-EME including ultrasound power density, temperature, applied voltage and extraction time were optimized with haloperidol, fluoxetine, and sertraline as model analytes. Under the optimal conditions, extraction recoveries of model analytes in water samples were in the range of 66.8%–91.6%. When this UA-EME method was coupled with LC-MS/MS for detection of the target analytes in human urine samples, the linear range of the analytical method was 10–1000 ng mL⁻¹, with R² > 0.997 for all analytes. The limits of detection (LOD) and limits of quantification (LOQ) were in the range of 1.7–2.1 ng mL⁻¹ and 5.7–6.7 ng mL⁻¹, respectively. The UA-EME expands the application field of ultrasound chemistry and will be very important in development of stable and fast sample preparation systems in the future.