Objectives: The mechanism of matrix metalloproteinases (MMPs) and their tissue inhibitors (TIMPs) in diabetic foot ulcers (DFUs) is unclear. The purpose of this study was to describe changes in MMP-1, MMP-9, and TIMP-1 levels during DFU healing, and to search for any correlation in the changes in MMP levels with wound healing, in order to find possible predictors of healing. Methods: Patients with a DFU were recruited and placed into two groups, according to the degree of wound healing: ‘good healers’ and ‘poor healers’. Levels of MMP-1, MMP-9, and TIMP-1 were analysed by ELISA (enzyme-linked immunosorbent assay). Results: A total of 22 patients participated in the study. The MMP-1 level was significantly higher at weeks zero (W0) and 12 (W12) in ‘good healers’ than in ‘poor healers’ (p=0.045 and 0.008, respectively). In contrast, the MMP-9 level was significantly lower in ‘good healers’ than in ‘poor healers’ at W0, W4, and W12 (p=0.001, 0.001 and 0.028, respectively). Receiver operator curve (ROC) analysis of the MMP-9 level, MMP-1/TIMP-1 ratio, and MMP-9/TIMP-1 ratio at W0 provided cut-off levels of 0.38, 0.056, and 9.06, respectively, which were best predictive of a reduction in wound area at W4 (‘good healers’ versus ‘poor healers’; thereby predicting wound healing condition at W12) with a sensitivity of 81.8%, 81.8%, and 90.9%, and a specificity of 64.6%, 55%, and 64.6%, respectively. Conclusion: A ‘poor healing scoring system’ is therefore proposed that could be determined on patient admission, which has the potential to be used clinically as a predictor of healing, thus allowing an appropriate treatment plan to be developed.