Maize (Zea mays L.) grain yield is closely associated with kernel number at harvest. This yield component is a function of the physiological condition of the crop at a critical period bracketing silking. The objectives of this study were to analyze for that critical period by assessing (i) the relationship between final kernel number and plant or crop growth rate, (ii) the partitioning of dry matter to reproductive structures, and (iii) the effect of duration of this period on kernel set. Kernel number (per plant and per unit area) and growth rates of crops, plants, and ears were measured in 12 field experiments conducted at Balcarce, Argentina, from 1989 to 1996 with hybrid Dekalb 636. Experimental sources of variation included shading, plant density, sowing date, and night temperature. Shading and night temperature treatments were established during a 31‐d period bracketing silking. In one of the density experiments, plant growth rate was estimated for individual plants by means of a non‐destructive methodology based on correlations between dry matter and morphometric measurements. The relationship between kernel number per plant and plant growth rate was curvilinear with threshold values of approximately 1 g d−1 for kernel set and 6 g d−1 for prolificacy. At high and at low plant growth rates during the period bracketing silking, Dekalb 636 set a low number of kernels per unit of plant growth rate. This effect was explained by limited partitioning of biomass to reproductive structures during that period. When duration of the critical period for kernel set was considered by expressing plant growth rate in thermal units, a better estimation of kernel number per plant was obtained.