tAn increasing demand for lightweight casual garments has driven the need for cotton with lower fibrefineness to facilitate production of finer yarns in the spinning mill. Breeding cottons with finer fibre(lower linear density) to address this target will reduce lint yield unless there is a concomitant changein another yield component. One yield component to consider is seed fibre density (FD) which is thenumber of fibres per unit seed surface area (SSA). FD can be measured also as the number of fibre initialson the seed coat at fibre initiation by scanning electron microscopy (SEM). In order for a cotton breederto apply selection during breeding, an accurate measure of FD is required.Micronaire is a common, indirect measurement of cotton fineness but it is confounded by fibre maturity.This paper compares the use of micronaire with a direct measure of fineness in calculating FD. It comparesmature seed estimated FD to FD determined by SEM imagery of one day old ovules, and determines themost reliable stage during flowering for quantifying FD. Data from five field experiments provided 1469samples in which micronaire and fineness were compared for estimating FD. When using fibre finenessfor calculation, the average FPS was 17,017, and the average FD was 157 mm−2. The average FD by SEMwas 5108 mm−2, it was higher than FDfin because ovules were ∼4% of the final seed surface area due togrowth dilution. Although micronaire and fineness were correlated, there was scatter in that relationshipdue to differences in fibre maturity. As a result, there were many instances where FD was overestimatedwhen calculated by micronaire, compared with using fineness, such that less than 50% of the top 20% FDmeasured by fineness were identified by the top 20% of the samples in FD measured by micronaire. Resultsindicated FD measured by fineness on mature seeds obtained similar ranking to direct measures by SEMof FD on one day old ovules and the mid flowering stage was more reliable than early or late flowering. Itwas concluded that FD should be estimated by using fineness rather than micronaire because micronaireis not a reliable measure of fineness.