To describe in detail the arterial vasculature of metacarpophalangeal joints 2-5 on cadaver specimens and to compare it to ultrasound imaging of healthy subjects. Eighteen hands of donated human cadavers were arterially injected and investigated with either corrosion casting or cryosectioning. Each layer of cryosectioned specimens was photographed in high-resolution. Images were then segmented for arterial vessels of the metacarpophalangeal (MCP) joints 2-5. The arterial pattern of the joints was reconstructed from the segmented images and from the corrosion cast specimens. Both hands of ten adult healthy volunteers were scanned focusing on the vasculature of the same joints with high-end ultrasound imaging, including color Doppler. Measurements were made on both cryosectioned arteries and Doppler images. The arterial supply of MCP joints 2-5 divides into a metacarpal and a phalangeal territory, respectively. The metacarpal half receives arteries from the palmar metacarpal arteries or proper palmar digital arteries, while the phalangeal half is supplied by both proper and common palmar digital arteries. Comparing anatomical and ultrasonographic results, we determined the exact anatomic location of normal vessels using Doppler images acquired of healthy joints. All, except three branches, were found with less than 50% frequency using ultrasound. Doppler signals were identified significantly more frequently in MCP joints 2-3 than on 4-5 (p < 0.0001). Similarly, Doppler signals differed in the number of detectable small, intraarticular vessels (p < 0.009), but not that of the large extraarticular ones (p < 0.1373). When comparing measurements acquired by ultrasound and on cadaver vessels, measurements using the former technique were found to be larger in all joints (p < 0.0001). Using morphological and ultrasonographic techniques, our study provides a high-resolution anatomical maps and an essential reference data set on the entire arterial vasculature of healthy human MCP 2-5 joints. We found that Doppler signal could be detected in less than 50% of the vessels of healthy volunteers except three locations. Intraarticular branches were detected with ultrasound imaging significantly more frequently on healthy MCP 2-3 joints, which should be taken into account when inflammatory and normal Doppler signals are evaluated. Our study also provides reference data for future, higher-resolution imaging techniques.