A controlled feeding experiment, designed to maintain liveweight while changing dietary crude protein content, was conducted with medium wool Merino wethers of high (fleece-plus) and low (fleece-minus) wool growing capacity. The sheep were split into 3 groups and offered a ration of medium (17%) protein content throughout the experiment (Control), a sequence in which the ration changed from medium to high (27%) then low (7%) protein content, or a ration sequence of medium'low'high protein content. While diets were offered at a level designed to be isoenergetic, incomplete intake of the low protein ration resulted in the energy intake of sheep receiving this ration being significantly (P < 0.05) less than the energy intake of sheep receiving the other two rations. Feeding regime significantly influenced staple strength of fleece-plus, but not fleece-minus sheep. Fleece-plus sheep receiving the ration sequence that changed from medium to low to high protein content grew wool with significantly (P < 0.05) lower staple strength than fleece-plus sheep receiving either of the other ration sequences. Liveweight changes throughout the experiment were small and did not explain any of the variance in staple strength. Analysis of fibre diameter profiles measured using autoradiography indicated along-fibre variation in diameter was influenced by genotype and feeding sequence, while between-fibre variation in diameter was influenced by feeding sequence only. Regression analysis indicated two models, one containing between-fibre variation in diameter and standard deviation of fibre diameter, and another including along-fibre variation in diameter, rate of fibre diameter increase during ration changes, and clean fleece weight explained a significant part of the variation in staple strength. The results are discussed in relation to mechanisms affecting staple strength.