Magnetocardiography (MCG) has been introduced as an innovative non-invasive diagnostic tool to identify various heart diseases. However, there have been little data on the reliability of MCG parameters. The purpose of this study is to examine the test–retest reliability of different diagnostic parameters derived from MCG. We investigated short-, intermediate-, and long-term reliability of nine parameters from T max/3–T max interval, and five parameters from each time point such as QRS-wave, the peak of R-, and T-wave were evaluated. Short-term reliability was tested in the youngest 20 subjects (mean age = 26.3 ± 4.9 years) in three sessions separated by 5 min. Intermediate-term reliability was tested in the 35 subjects with coronary artery disease (CAD) (65.1 ± 7.1 years) with two recording sessions each in the morning and afternoon, separated by more than four hours. Long-term reliability was tested in seven subjects (37.1 ± 8.8 years) using seven daily sessions. Interclass correlation coefficients (ICC) showed that test–retest reliability was good to excellent (0.99 ≥ ICC ≥ 0.80) for six out of nine parameters within T max/3–T max. In addition, all parameters on the peak of R-wave, T-wave, and QRS-wave integrated were good to excellent (0.99 ≥ ICC ≥ 0.80) except for one parameter of CAD patients showing lower ICC values under 0.7. In conclusion, our study showed that the test–retest characteristics of the studied MCG parameters are generally stable and reliable over periods of minutes to days in subjects with different age spectrums.