Bending and compressive loads were applied intermittently at 0.5 Hz through implants chronically inserted into the radius and ulna of experimental sheep. The plane of bending was the same as that imposed during locomotion. The principal variables of the artificial strain regime were the peak strains and strain rates, which never exceeded the range attainable during normal locomotion. The bones were loaded artificially for one hour per day for six weeks. Post mortem, the parameters of the bones' remodelling response were assessed by measuring the change in gross geometry and the amount of new bone formation which was marked with fluorescent label given during the experimental period. The strain-related variable, which had the greatest influence on every remodelling parameter investigated, was the ratio between the maximum strain rate of the artificial regime and the maximum strain rate during walking. The variation in this ratio could be shown to account for between 68 and 81% of the variation in the measures of surface bone deposited. The strain parameter which caused the greatest additional increase in the proportion of total variation explained was most commonly the ratio between the proportion of peak strain due to axial loading in the normal and artificial situations. The effect of this axial strain was far less marked than that of strain rate, however, and only increased the percentage of the total variance which could be explained by between 6 and 12%. The direction of bending and axial loading (tension or compression) appeared to have no effect on the course of the remodelling observed. The most effective influence on the amount of intracortical secondary osteal remodelling was also the maximum strain rate ratio. Its effect on this process seemed less marked than that on surface remodelling, however, since it could only explain 43% of the variance in the total number of secondary osteons formed. The addition of other strain-related variables did not significantly increase this proportion.