Cotton stickiness is a serions problem for many producing countries. In Sudan, 1000 bales of the 1996 production WERE selected and sampled to study the féasibility of a bale stickiness classification. Sixteen thousand samples tested on the Sticky Cotton Thermodetector show that within-bale variance increases with the mean. Statistical investigations show that the distribution of sticky points for certain bales is over-dispersed with respect to the Poisson distribution. Fitting of' Negative binomial distribution was studied, but the dispersion parameter seems to be dependent on the mean. This law will therefore be difficult to apply.