A feasibility study of microplastic detection and quantification in soil and water using resonance microwave reflectometry is carried out using artificially created samples with a high volumetric concentration of microplastic with 50μm-0.5mm particles size. A mathematical model expressing microplastic concentration in soil and water as a linear function of the measured S11 resonance frequency shift and relative permittivity contrast is developed and is found to be in an excellent agreement with the experimental data based on synthetic contaminated material samples. Next, this model is applied to find the best achievable theoretical resolution of microplastic concentration in the natural environment using microwave sensing technology, which is shown to be at around 100ppm (parts-per-million) level in the linear signal detection regime. It is demonstrated that the best achievable level of microplastic contaminant resolution depends on the sensor probe Q-factor and sensitivity of the microwave receiver. The bound for the achievable contaminant concentration resolution is found in the analytical form for high-Q resonance microwave sensors of arbitrary geometry. Even though several well-established protocols based on optical, infrared, and X-ray spectroscopy are currently being used for microplastic detection in the natural environment, microwave spectroscopy could offer additional benefits, especially for low-cost, real-time in-situ microplastic detection in diverse environmental conditions outside of the laboratory space.