Applying the Classical Nucleation Theory (CNT) to crystallization of glasses encounters some difficulties. One of the most important aspects is that this theory overlooks structural relaxation by assuming that crystal nucleation proceeds in a relaxed, metastable, supercooled liquid (SCL). Considering this assumption, the thermodynamic driving force, diffusion coefficient, and surface tension should be constant at any given temperature. Here, we performed experiments for very extended times (up to about 2,200 hours at 703K), at temperatures below the laboratory glass transition, Tg=726K, of a lithium disilicate glass used as a model. Our results show that crystal nucleation starts concomitantly with the relaxation process of the glass towards the SCL, which strongly affects the nucleation kinetics, taking over 500 hours to reach the ultimate steady-state regime at this temperature. This very long relaxation process proceeds much slower than the well-known alpha-relaxation determining, e.g., the temporal evolution of the glass density, which takes only ~20 hours at this same temperature. Nevertheless, structural relaxation results in a decrease of the work of critical cluster formation leading to an upsurge of the nucleation rate. The increased nucleation rate mainly reflects this long structural relaxation mode of the glass and is not related to the classical transient nucleation, which has been exclusively employed so far in the interpretation of nucleation kinetics by most researchers, including ourselves, over the past 40 years. Our experimental results and analyses prove effect of glass relaxation on crystal nucleation and shed light on the alleged “breakdown” of the CNT at low temperatures.