Matrix effect-free UHPLC–MS/MS method was developed and validated for the determination of cholesterol-lowering lovastatin in food samples represented by Pu-erh tea, oyster mushroom, and red yeast rice. The resulting method was fully validated in terms of intra-day and inter-day precision, accuracy, linearity, range, LOD, LOQ, and matrix effects. The matrix effect phenomenon evaluated by comparison of slopes of calibration curves was completely eliminated by solid-phase extraction based on the technique of molecularly imprinted polymers (MIPs). Comparison of elution profiles obtained on the MIP and corresponding control non-imprinted polymer (NIP) showed selectivity of the extraction procedure. In addition, selectivity of the MIP material and the molecularly imprinted solid-phase extraction (MISPE) was also proved by experiments evaluating retention of analytes physico-chemically similar to the target molecule. Extraction recoveries of these analytes represented by estrogen derivatives (estrone, estriol, 17α-ethinylestradiol, and β-estradiol) were very low or even null. Synthesis and preparation of the resulting MIP sorbent was characterized by excellent repeatability expressed as RSD 7.7% (n = 9) of extraction recoveries. The determined capacity of the MIP material reaching 375 ng/mg is sufficient for analysis of the evaluated statin in its natural sources. Suitability of the resulting MISPE-UHPLC–MS/MS procedure for real sample analysis was verified by the determination of lovastatin in one dietary supplement based on the red yeast rice with a given amount of the target analyte. Finally, three mushroom and fifteen tea samples obtained in Czech food stores and tearooms were subjected to analysis. Low or null amount of lovastatin was found in these samples.