We present results of extensive theoretical and experimental investigations of Mg2Si and Mg2Sn and their Mg2Si1-xSnx alloys. Electronic and phonon properties of binary compounds were studied by ab initio calculations. Then, both compounds were synthesized by the solid-state reaction and electrical resistivity and thermopower was measured at high temperature (300–900 K). In both the compounds, the theoretical bandgaps (0.56 eV in Mg2Si and 0.16 eV in Mg2Sn) agree very well with the experimental values (0.6 eV in Mg2Si and 0.17 eV from activation law in Mg2Sn) upon applying the modified Becke–Johnson semilocal exchange potential and including spin–orbit coupling in the calculations. Calculated phonon spectra support crystal stability of both compounds. For Mg2Si, the contributions from Si and Mg are spread over all the spectrum (0–10 THz), whereas in the case of Mg2Sn, a gap opens around 4 THz with Sn and Mg contributions dominating in lower and higher energy range, respectively. The calculated heat capacity at low temperature (0–300 K) fairly agrees with available experimental data. The crystal structure of Mg2Si1-xSnx with x = 0, 0.25, 0.4, 0.75, 1 was studied by X-ray diffraction measurements. The alloy compositions exist in the ranges 0 < x < 0.4 and 0.6 < x < 1 and the obtained samples are almost single phased. Detailed crystal stability study with temperature revealed that all powder samples started to decompose into MgO, Si and Sn at ~ 630 K. For hot pressed bulk materials, the decomposition is much slower than in powder compounds but it still appears. Interestingly, thermoelectric properties measurements performed in Mg2Si1-xSnx show that both electrical resistivity and thermopower curves are repeatable during temperature cycles up to 770 K. On the other hand, temperature-dependent X-ray powder diffraction suggests that these compounds are not stable. Furthermore, electronic structure calculations of almost 40 impurities (s- and p-block, 3d and 4d transition metal elements) diluted in Mg2Si and Mg2Si0.75Sn0.25 were performed by the KKR-CPA method. Based on calculated impurity density of states the character of doping (n or p) is predicted, which, however, strongly depends on the substituted crystallographic site.