The balance between self-renewal and differentiation of neural progenitor cells (NPCs) dictates neurogenesis and proper brain development. We found that the RNA- binding protein Sam68 (Khdrbs1) is strongly expressed in neurogenic areas of the neocortex and supports the self-renewing potential of mouse NPCs. Knockout of Khdrbs1 constricted the pool of proliferating NPCs by accelerating their cell cycle exit and differentiation into post-mitotic neurons. Sam68 function was linked to regulation of Aldh1a3 pre-mRNA 3'-end processing. Binding of Sam68 to an intronic polyadenylation site prevents its recognition and premature transcript termination, favoring expression of a functional enzyme. The lower ALDH1A3 expression and activity in Khdrbs1-/- NPCs results in reduced glycolysis and clonogenicity, thus depleting the embryonic NPC pool and limiting cortical expansion. Our study identifies Sam68 as a key regulator of NPC self-renewal and establishes a novel link between modulation of ALDH1A3 expression and maintenance of high glycolytic metabolism in the developing cortex. DOI: http://dx.doi.org/10.7554/eLife.20750.001, eLife digest Neurons develop from cells called neural progenitors. These cells can either divide to produce more progenitor cells or develop into specific types of neurons. These two activities – known as self-renewal and differentiation – must be balanced to produce the right number of specialized neurons, without depleting the pool of progenitor cells. The self-renewal and differentiation of progenitor cells is balanced by essentially regulating which genes are active, or expressed, within the cells. In the first step of gene expression, the genetic instructions are copied to form a molecule of pre-messenger RNA (or pre-mRNA for short). Each pre-mRNA molecule is then processed to produce a final product that can be translated into protein. Importantly, two copies of the same pre-mRNA may sometimes be processed in different ways, which allows multiple proteins to be produced from a single gene. RNA-binding proteins control pre-mRNA processing. The expression of one such protein, called Sam68, oscillates during the development of the nervous system, such that its expression peaks when there is intense production of new neurons and then declines. However, it was not known whether Sam68 actually helps neurons to develop. La Rosa et al. have now analysed the role of Sam68 in the developing brain of mice. The experiments confirmed that Sam68 is highly expressed in neural progenitor cells and showed that its levels dictate the cell’s fate: high expression encourages a cell to self-renew, while low expression triggers it to develop into a specialized neuron. Further investigation revealed that Sam68 works by promoting the expression of a metabolic enzyme called Aldehyde Dehydrogenase 1A3 or ALDH1A3. This enzyme promotes the release of energy from molecules of glucose via a process known as anaerobic glycolysis. La Rosa et al. found that cells that lack Sam68 make a truncated version of the pre-mRNA encoding ALDH1A3. This truncated pre-mRNA encodes a shortened version of the enzyme that is inactive. Further experiments confirmed that Sam68 normally prevents this from happening by binding to the pre-mRNA and processing it to produce the full-length, working version of the ALDH1A3 enzyme. Also, La Rosa et al. found that progenitor cells need working ALDH1A3 to keep them dividing, and to stop them from developing into specialized neurons too soon. Finally, because the processing of pre-RNA plays a major role in brain development, problems with this process often lead to intellectual disabilities and neurodegenerative diseases, such as autism spectrum disorder and amyotrophic lateral sclerosis. The next step following on from these new findings will be to investigate whether defects in Sam68 contribute to such conditions and, if so, to look for ways to counteract these defects. DOI: http://dx.doi.org/10.7554/eLife.20750.002