This article critically reviews some ethanol fermentation technologies from sugar and starch feedstocks, particularly those key aspects that have been neglected or misunderstood. Compared with Saccharomyces cerevisiae, the ethanol yield and productivity of Zymomonas mobilis are higher, because less biomass is produced and a higher metabolic rate of glucose is maintained through its special Entner–Doudoroff pathway. However, due to its specific substrate spectrum as well as the undesirability of its biomass to be used as animal feed, this species cannot readily replace S. cerevisiae in ethanol production. The steady state kinetic models developed for continuous ethanol fermentations show some discrepancies, making them unsuitable for predicting and optimizing the industrial processes. The dynamic behavior of the continuous ethanol fermentation under high gravity or very high gravity conditions has been neglected, which needs to be addressed in order to further increase the final ethanol concentration and save the energy consumption. Ethanol is a typical primary metabolite whose production is tightly coupled with the growth of yeast cells, indicating yeast must be produced as a co-product. Technically, the immobilization of yeast cells by supporting materials, particularly by gel entrapments, is not desirable for ethanol production, because not only is the growth of the yeast cells restrained, but also the slowly growing yeast cells are difficult to be removed from the systems. Moreover, the additional cost from the consumption of the supporting materials, the potential contamination of some supporting materials to the quality of the co-product animal feed, and the difficulty in the microbial contamination control all make the immobilized yeast cells economically unacceptable. In contrast, the self-immobilization of yeast cells through their flocculation can effectively overcome these drawbacks.