Mastitis is a major concern in sheep farming. Detecting subclinical mastitis, which manifests without apparent signs, is particularly challenging. While somatic cell count (SCC) and the California Mastitis Test (CMT) are widely used in dairy animals, and infrared thermography has shown some promise, their diagnostic performance in meat sheep remains under-evaluated. This study aimed to 1) characterize SCC and udder skin surface temperature across lactation in subclinically infected and uninfected udder halves; 2) determine diagnostic cutoffs for SCC, CMT, and thermography; and 3) compare diagnostic performance using receiver operating characteristic curves. Milk and thermal data were collected weekly from 37 clinically healthy nursing ewes over the first 8 weeks postpartum. Infection was defined by ≥100 CFU/mL of a bacterial species. Infected udder halves had higher log SCCs than uninfected halves in weeks 1–4 and 6 postpartum. Infection status did not predict udder skin temperature, which was more influenced by ambient temperature. SCC had the highest diagnostic performance (AUC = 0.78), with an optimal cutoff of 148,500 cells/mL (sensitivity = 0.77; specificity = 0.66). The CMT yielded an AUC of 0.70 with an optimal cutoff score above negative (i.e., trace or positive reaction; sensitivity = 0.77; specificity = 0.51). Udder skin temperature had the lowest AUC (0.56). SCC is the most effective of the three tools for detecting subclinical mastitis, although CMT may serve as a useful on-farm screening tool. Infrared thermography did not detect subclinical infection, highlighting the need for further research into non-invasive diagnostics.