Ten cultivars of spring wheat (Triticum aestivum L.) were compared on the basis of grain yield determined at intervals after anthesis. When the data on grain yield were transformed to a square root scale, regression methods showed that the growth curves of the grain weight could be adequately represented by low-order polynomial expressions of time from anthesis. The fitted regression coefficients allowed the mid-season filling rate (and its standard error) to be estimated for each variety, and consequently the varietal filling rates could be compared. While the rate of filling appeared to be an important factor in determining the final weight, the length of the grain filling period (anthesis to maturity), by itself, was not.