PalmElit implements the genetic improvement and marketing programs for CIRAD® oil palm seeds. The commercial seeds embody 80 years of genetic improvement work undertaken by IRHO, CIRAD and PalmElit in conjunction with several partners of excellence located on each of the continents where oil palm is grown. An increase of more than 60% in oil yields was achieved since 1960. This result illustrates the efficiency of the recurrent reciprocal selection (RRS) underlying the conducted breeding program. So far, assessment of parental breeding values has largely relied on progeny testing, which is an efficient but time- and money-consuming step within the RRS scheme. With the recent development of oil palm genomic resources, genomic selection (GS) appears as an attractive strategy to increase the efficiency of oil palm breeding programs. On a theoretical point of view, GS has the potential to increase the rate of genetic gain by shortening the breeding cycle and/or increasing the selection intensity. PalmElit, together with its research partner CIRAD, has been leading research for nearly 10 years in order to develop and assess the implementation of GS in oil palm breeding. Some of the key achievements have been shared with the scientific community since 2015 (Cros et al., 2015a, 2015b, 2017a; Marchal et al., 2016) which corroborate the potential of GS in terms of increased genetic gain. Further research is still ongoing to answer the simple -but critical- question: what is the optimal use of GS in terms of genetic gain vs time- and cost-efficiency? In this paper, following a brief review on the GS history and key concepts, we present our latest results which address critical aspects such as prediction accuracy and optimal use of GS within breeding schemes. We extend and discuss our conclusions in light of the literature available in oil palm and other crop species. Finally, we summarize the perspectives and challenges for successful implementation of GS in oil palm.