Starved horses with low body condition scores (BCS, < 3 out of 9) continues to be a prominent welfare concern, yet limited nutritional studies on the reconditioning process exist. The aim of this study was to determine how alternative diets affected blood chemistry panels and body composition in mature, starved (initial BCS 1 or 2) light breed horses during the refeeding process (final BCS 4, UTK IACUC 2811–1220). Horses (n = 10, 18.8 ± 2.5 y (mean ± SD), BCS 1.4 ± 0.5, 3 geldings and 6 mares, light breeds) were placed on a control diet (C; 100% digestible energy, DE, from grass hay and Purina Free Balance) or a concentrate diet (S; 50 % DE from grass hay and 50% DE from Purina Equine Senior). Diets were formulated to increase caloric intake in accordance with previously established refeeding recommendations and the NRC (2007), where horses were increased from 50% DE to 100% DE over 10 d to prevent refeeding syndrome, and then were increased up to 145% DE as needed based on weekly body weight (BW) until a BCS of 4 was reached. Weekly BW and BCS were assessed by 3 independent reviewers and plasma samples collected at each BCS increase based on reviewer average and an increase in of 16–20 kg in BW. Diet and BCS effects on blood chemistry and body composition were analyzed using generalized linear mixed models in SAS, with breed and individual animals as random effects, and mean separation using Tukey's adjustment.Significance was detected at P < 0.05 (SAS 9.4, Cary, NC). Non-normally distributed data were rank transformed as determined by Shapiro-Wilk. Of the 10 horses in the study, 3 were euthanized within 10d of enrollment due to infectious disease or colic unassociated with dietary treatment, 1 was disenrolled due to pneumonia, and the remaining 6 horses were fully reconditioned (C: n = 3, S: n = 3). Horses were recovered similarly between diet groups (change in BW: C = 46.15 ± 9.9 kg, S = 41.8 ± 14.9 kg, mean ± SD, P > 0.05). Plasma creatine kinase differed by BCS (P = 0.0007), whereas creatinine differed by both diet (P = 0.006) and BCS (P < 0.0001); however, all values were within normal limits. Lactose was lower in BCS 4 horses compared with all other scores, yet all values were also within normal ranges (P = 0.0007). Rib fat thickness was greater in BCS 4 horses (C = 0.35(0.08)mm, S = 0.38(0.20)mm, median(IQR)) than all other BCS (P = 0.001). Rump fat thickness was highest in BCS 4 horses (S = 0.36 (0.13), C = 0.30 (0.05)). Based on our findings, grass hay and complete feed concentrates can be used to successfully recover emaciated horses. Further studies would be helpful in determining optimal refeeding protocols.