Feed intake affects the GH-IGF system and may be a key factor in determining the ovarian follicular growth rate. In fat mares, the plasma IGF-1 concentration is high with low GH and a quick follicular growth rate, in contrast to values observed in thin mares. Nothing is known regarding the long-term effects of differential feed intake on the IGF system. The objective of this experiment was to quantify IGFs, IGFBPs, GH, glucose, insulin, gonadotropin and progesterone (P4) in blood and in preovulatory follicular fluid (FF) in relation to feeding levels in mares. Three years prior to the experiment, Welsh Pony mares were assigned to a restricted diet group (R, n = 10) or a well-fed group (WF, n = 9). All mares were in good health and exhibited differences in body weight and subcutaneous fat thickness. Follicular development was scanned daily and plasma was also collected daily. Preovulatory FF was collected by ultrasound-guided follicular aspiration. Hormone levels were assayed in FF and plasma with a validated RIA. According to scans, the total number of follicles in group R was 53% lower than group WF. Insulin and IGF-1 concentrations were higher in WF than in R mares. GH and IGF-2 concentrations were lower in plasma from WF mares than from R mares, but the difference was not significant in FF. The IGFBP-2/IGFBP-3 ratio in FF was not affected by feeding but was dramatically increased in R mare plasma. No difference in gonadotropin concentration was found with the exception of FSH, which was higher in the plasma of R mares. On the day of puncture, P4 concentrations were not affected by feeding but were higher in preovulatory FF than in plasma. The bioavailability of IGF-1 or IGF-2, represented by the IGFBP2/IGFBP3 ratio, is modified by feed intake in plasma but not in FF. These differences partially explain the variability in follicular growth observed between well-fed mares and mares on restricted diets.