Objective Clinical heterogeneity of primary osteoarthritis (OA) is a major challenge in understanding pathogenesis and development of targeted therapeutic strategies. This study aims to (1) identify OA patient subgroups phenotypes and (2) determine predictors of OA severity and cartilage-derived stem/progenitor concentration using clinical-, tissue-, and cell- level metrics. Design Cartilage, synovium (SYN) and infrapatellar fatpad (IPFP) were collected from 90 total knee arthroplasty patients. Clinical metrics (patient demographics, radiograph-based joint space width (JSW), Kellgren and Lawrence score (KL)), tissue metrics (cartilage histopathology grade, glycosaminoglycans (GAGs)) and cell-based metrics (cartilage-, SYN-, and IPFP-derived cell concentration ([Cell], cells/mg), connective tissue progenitor (CTP) prevalence (PCTP, CTPs/million cells plated), CTP concentration, [CTP], CTPs/mg)) were assessed using k-mean clustering and linear regression model. Results Four patient subgroups were identified. Clusters 1 and 2 comprised of younger, high body mass index (BMI) patients with healthier cartilage, where Cluster 1 had high CTP in cartilage, SYN, and IPFP, and Cluster 2 had low [CTP] in cartilage, SYN, and IPFP. Clusters 3 and 4 comprised of older, low BMI patients with diseased cartilage where Cluster 3 had low [CTP] in SYN, IPFP but high [CTP] in cartilage, and Cluster 4 had high [CTP] in SYN, IPFP but low [CTP] in cartilage. Age ( r = 0.23, P = 0.026), JSW ( r = 0.28, P = 0.007), KL ( r = 0.26, P = 0.012), GAG/mg cartilage tissue ( r = −0.31, P = 0.007), and SYN-derived [Cell] ( r = 0.25, P = 0.049) were weak but significant predictors of OA severity. Cartilage-derived [Cell] ( r = 0.38, P < 0.001) and PCTP ( r = 0.9, P < 0.001) were moderate/strong predictors of cartilage-derived [CTP]. Conclusion Initial findings suggests the presence of OA patient subgroups that could define opportunities for more targeted patient-specific approaches to prevention and treatment.