Cheese supplies bioactive peptides, fatty acids (FA), minerals, and vitamins essential for human health. Common laboratory analyses of these components are expensive and time consuming. Near-infrared spectroscopy is a rapid, objective, non-destructive, and cheap method to determine several composition traits. However, heterogeneity of cheese, and low concentration of FA and minerals make their prediction difficult. This study aimed to develop prediction models for cholesterol, FA profile, and mineral content of commercial European cheeses using near infrared transmittance (NIT) spectroscopy. A total of 145 ground cheese samples from different dairy species and ripening time (fresh to 24 mo) were scanned with a NIT spectrophotometer every 2 nm from 850 to 1,050 nm wavelength. Sample spectra were matched with absolute content of cholesterol, FA, and mineral reference data to develop prediction models. Modified partial least squares regressions were validated through external validation after dividing the data in calibration (75%) and external validation (25%) sets. Cheese moisture, fat, protein, total solids, and cholesterol averaged 43.24 ± 0.97%, 27.24 ± 0.47%, 24.87 ± 0.54%, 56.76 ± 0.97%, and 0.07 ± 0.001%, respectively. Cholesterol content was inadequately predicted, exhibiting a coefficient of determination of external validation (R²_ExV) of 0.50 and a residual prediction deviation of external validation (RPD_ExV) of 1.36. Satisfactory models were developed for saturated, unsaturated, monounsaturated, and polyunsaturated FA, and myristic, palmitic, oleic, and some minor FA (R²_ExV from 0.87 to 0.97; RPD_ExV from 2.74 to 4.73). Promising predictions were obtained for Ca, Na, P, S, Mg, Zn, and Cu (R²_ExV from −0.94 to 0.83; RPD_ExV from −3.73 to 2.35). Results of the present study are a prelude to the at-line utilization of prediction models for the most abundant cheese FA and minerals.