The biological mechanisms underlying differences in wool staple strength were examined in 2 groups of Merino sheep that have been genetically selected for high or low staple strength, while holding fibre diameter constant. The sheep were fed below maintenance for 87 days, and then re-fed ad libitum for 63 days with a diet containing either 9% or 23% crude protein, in a cross-over design, after which they returned to the paddock until shearing. The fleeces of the 2 groups differed in staple strength (25·2 v.17·5 N/ktex, P < 0·001) but were similar in mean fibre diameter and clean fleece weight. However, the pattern of wool growth was different. When fed below maintenance, sheep from the sound group grew more wool than sheep from the tender group (P < 0·05), but lost more liveweight (P < 0·01). During re-feeding, the sound sheep grew less wool than the tender sheep (P < 0·05), but gained more liveweight. The mean fibre diameter at the point of break was similar in both groups. Immediately after re-feeding, the fibre diameter increased more rapidly in the tender group than in the sound (P < 0·001), but a similar difference was observed between the high and low protein diets, with no effect on staple strength. The sound sheep had a lower standard deviation of fibre diameter than the tender sheep (P < 0·001), both in the whole fleece and in 3-weekly midside patches. The data indicated that the variability of fibre diameter between fibres made a larger contribution than the variability along fibres to the difference between the groups in overall variability of fibre diameter. The sheep were then grazed together at pasture for a second year and again differed in staple strength. In addition, the sound sheep grew less wool on green spring pastures and had a lower clean fleece weight (P < 0·05). The differences in wool growth rates between sheep from the sound and tender lines depended more on whether pasture was green than on the amount of pasture available. We conclude that the difference in staple strength between the sound and tender groups was most closely associated with the variability between fibres in diameter, and was also affected by a difference in variation in diameter along the fibres. Staple strength was not affected by the amount of wool at the point of break, or by the rate of change in fibre diameter after feeding. The sheep in the sound group grew less wool than those in the tender group when on good nutrition.