The increasing relevance of cottonseed in a farmer's revenue makes it necessary to have reliable methods to measure its yield. There are two approaches to measuring cottonseed yield. Actual cottonseed yield, obtained from a grab-sample at the ginning mill, and estimated cottonseed yield, obtained from lint yield and lint percentage of seed cotton. These two measures may differ due to various factors. The purpose of this research was to estimate seed yield from lint yield and lint percentage of seed cotton, and to determine biasedness and accuracy of the estimate. A sample of 468 experimental trials, with three Texas locations, two irrigation levels, and 15 varieties was used in the statistical analysis, which consisted of three parts. The first part deter-mined that there is a 9.1 lb. per acre bias in cottonseed yield estimation. The second part determined that variety and irrigation level were significant sources of bias. For the irrigation level effect, dryland trials showed no significant bias, while irrigated trials showed a 21.5 lb. per acre bias. The third part of the analysis showed a close relationship between actual and estimated seed yields (r = .99), and a standard error of the estimate of 78.3 lb. per acre. It was concluded that estimated cottonseed yield (from lint yield and lint percentage of seed cotton) can be used as a reliable estimator of actual cottonseed yield.