Background. During transcranial electrical stimulation (tES), including transcranial direct current stimulation (tDCS) and transcranial alternating current stimulation (tACS), current density concentration around the electrode edges that is predicted by simplistic skin models does not match experimental observations of erythema, heating, or other adverse events. We hypothesized that enhancing models to include skin anatomical details, would alter predicted current patterns to align with experimental observations. Method. We develop a high-resolution multi-layer skin model (epidermis, dermis, and fat), with or without additional ultra-structures (hair follicles, sweat glands, and blood vessels). Current flow patterns across each layer and within ultra-structures were predicted using finite element methods considering a broad range of modeled tissue parameters including 78 combinations of skin layer conductivities (S m–1): epidermis (standard: 1.05 × 10−5; range: 1.05 × 10−6 to 0.465); dermis (standard: 0.23; range: 0.0023 to 23), fat (standard: 2 × 10−4; range: 0.02 to 2 × 10−5). The impact of each ultra-structures in isolation and combination was evaluated with varied basic geometries. An integrated final model is then developed. Results. Consistent with prior models, current flow through homogenous skin was annular (concentrated at the electrode edges). In multi-layer skin, reducing epidermis conductivity and/or increasing dermis conductivity decreased current near electrode edges, however no realistic tissue layer parameters produced non-annular current flow at both epidermis and dermis. Addition of just hair follicles, sweat glands, or blood vessels resulted in current peaks around each ultrastructure, irrespective of proximity to electrode edges. Addition of only sweat glands was the most effective approach in reducing overall current concentration near electrode edges. Representation of blood vessels resulted in a uniform current flow across the vascular network. Finally, we ran the first realistic model of current flow across the skin. Conclusion. We confirm prior models exhibiting current concentration near hair follicles or sweat glands, but also exhibit that an overall annular pattern of current flow remains for realistic tissue parameters. We model skin blood vessels for the first time and show that this robustly distributes current across the vascular network, consistent with experimental erythema patterns. Only a state-of-the-art precise model of skin current flow predicts lack of current concentration near electrode edges across all skin layers.