Actinide compounds are very intriguing objects for the quantum chemistry because, on the one hand, these compounds are of great scientific and technological interest and, on the other hand, quantitative first principle based modeling of their electronic structure is extremely difficult because of strong relativistic effects and complicated electron correlation pattern. The efficiency of high-level all-electron relativistic methods in applications to complex actinide systems of practical interest is questionable and more economical but sufficiently accurate approaches to the studies of such systems are preferable. Recently, generalized relativistic effective core potentials (GRECPs) have been generated for actinides to perform accurate calculations of electronic structure and properties of their compounds with moderate computational cost. The accuracy of different GRECP versions is analyzed in atomic calculations and their applications to molecular and cluster calculations are reviewed. The results are compared with available experimental data and other theoretical studies. (c) 2015 Wiley Periodicals, Inc.