Aim: To evaluate the metacarpophalangeal (MCP) joints (MCP2 and 5) in healthy subjects by ultrasonography (US) using a high frequency transducer (18 MHz) and to verify the interobserver agreement. Material and methods: We enrolled 50 healthy volunteers (37 women, age between 30-58 years, mean age 41.7 years, divided into 3 groups according to age: 30-39, 40-49, and 50-58 years). The subjects were successively evaluated by 4 rheumatologists: 2 experienced (team A) and 2 beginners (team B) in US. Seven dorsal and palmar longitudinal scanning positions and a supplementary scan for MCP cartilage were performed. The bone surface (erosions, osteophytes), the intra-articular content (synovial thickening and vascularization, 4 grade scale), and the aspect of the metacarpal head cartilage were analyzed. The anterior palmar recess was measured. The time for examination was recorded. Results: Erosions were detected in 7% of joints by team A and 2% by team B (p<0.05, kappa agreement 0.567) in subjects over 40 years. The agreement by team A in the detection of the erosions was very good (kappa value 0.83). A moderate positive correlation was obtained between the presence of erosions and age (r= 0.401, p=0.004). Osteophytes were identified only on the dorsal scan in subjects over 50 years (in 3.5% of joints team A, 1.5% team B, p>0.05, kappa value 0.421). No grade 1 synovitis was observed by team A but 4 joints with grade 1 synovitis were identified by team B (p<0.05) from the dorsal scan. The dimensions of the palmar recess had large distribution (MCP 2 between 0.55-1.3 mm; MCP 5 between 0.6-1.2 mm). No statistical significant differences were obtained when comparing the dimensions of the two hands or the values obtained in agegroups (all p>0.05). No statistical significant correlations were obtained between the dimensions of palmar recess and the body mass index or dominant hand (all p>0.05). No pathological findings were found in the examination of the metacarpal head cartilage. Power Doppler investigation found the presence of grade 1 signal in 2.5% joints by team A and 1.5% by team B (p>0.05) only in the dorsal scans. The mean time for examination was 7.8±1.74 min in team A and 13.78±2.96 min in team B (p<0.05). Conclusions: In healthy subjects pathological findings are occasionally encountered, especially erosions and osteophytes. Using an 18 MHz transducer the aspect of grade 1 synovitis was not encountered in healthy non-inflammatory MCP joints. There is a permanent need for standardized training and examination in musculoskeletal US.