Cotton short fiber content (SFC) was estimated from the length statistics of longer fibers as measured by the Advanced Fiber Information System. Totals of 25,000 fibers from each of 15 calibration cottons were measured by AFIS. Regression analyses between SFC and mean length, coefficient of variation of length, and upper quartile length were performed. In order to investigate the impact of a biased fiber-sampling technique, a computer simulation using actual fiber length distributions from AFIS was devised. The computer program was used to electronically truncate the fibers by a normal distribution of length reduction having mean values of 1/8", 2/8", or 3/8", at values of CV = 15% and 30% each. Additional regressions were performed on the electronically modified data to study the degradation of accuracy and precision of SFC resulting from the simulated bias in the sampling. The authors conclude that SFC is highly correlated with other fiber length parameters (especially the variance of fiber length), and that the long-fiber parameters can be used to estimate or predict SFC measurements. The fiber length parameters from virtually truncated fiber distributions also predicted SFC with relatively good agreement. The shorter the truncation length and the lower the variation, the better the predicted results. This indicates that short fiber content can be estimated from a biased sample (such as measured by HVI) with an agreement depending on the mean length and its variation for fibers hidden in the comb (clamp), and that an improved sampling method for HVI should increase the accuracy of measured short fiber content.