Granitic rocks play special role in the dynamics and evolution of the Earth and its thermal regime. First, their compositional variability, reflected in the distribution of concentrations of radiogenic elements, provides constraints on global differentiation processes and large scale planetary evolution, where emplacement of granites is considered a particularly important process for the formation of continental crust. Second, heat production by radioactive decay is among the main heat sources in the Earth. Therefore knowledge of heat production in granitic rocks is pivotal for thermal modelling of the continental lithosphere, given that most radiogenic elements are concentrated in granitic rocks of the upper continental crust whereas heat production in rocks of the lower crust and lithospheric mantle is negligible. We present and analyze a new global database GRANITE2017 (with about 500 entries) on the abundances of heat producing elements (Th, U, K) and heat production in granitic rocks based on all available published data. Statistical analysis of the data shows a huge scatter in all parameters, but the following conclusions can be made. (i) Bulk heat production in granitic rocks of all ages is ca. 2.0 μW/m³. It is very low in Archean-Early Proterozoic granitic rocks (1.67 ± 1.49 and 1.25 ± 0.83 μW/m³, respectively) and there is a remarkable peak in heat production in Middle Proterozoic granites (presently 4.36 ± 2.17 μW/m³) followed by a gradual decrease towards Cenozoic granites (3.09 ± 1.62 μW/m³). Low heat production in the ancient continental crust may be important for preservation of cratonic lithosphere. (ii) There is no systematic correlation between the tectonically controlled granite-type and bulk heat production, although A-type (anorogenic) granites are the most radioactive, and many of them were emplaced in Middle Proterozoic. (iii) There is no systematic correlation between heat flow and concentrations of radiogenic elements. (iv) The present-day global average Th/U value is 4.75 ± 4.27 with a maximum in Archean-Early Proterozoic granites (5.75 ± 5.96) and a minimum in Middle-Late Proterozoic granites (3.78 ± 2.69). The Th/U ratio at the time of granite emplacement has a minimum in Archean (2.78). (v) The present-day K/U ratio is close to a global estimate for the continental crust only for the entire dataset (1.46 ± 1.63) × 10⁴, but differs from the global ratio for each geological time, and all anomalously high values are observed only in Archean-Early Proterozoic granites. (vi) We do not observe a systematic difference in radiogenic heat production between Archean and post-Archean granites, but rather recognize a sharp change in radiogenic concentrations and ratios from the Early Proterozoic to Middle Proterozoic granites. The Proterozoic anomaly may be caused by major plate reorganizations possibly related to the supercontinent cycle when changes in the granite forming processes may be expected, or it may even indicate a change in global thermal regime, mantle dynamics and plate tectonics styles. (vii) Our results provide strong evidence that secular change in the Urey ratio was not monotonous, and that plate motions may have been the fastest in Middle Proterozoic and have been decreasing since then. (viii) We estimate the total present-day heat production in the granitic crust as 5.8–6.8 TW and in the continental crust as 7.8–8.8 TW.