Red blood cells play a major role in body metabolism by supplying oxygen from the microvasculature to different organs and tissues. Understanding blood flow properties in microcirculation is an essential step towards elucidating fundamental and practical issues. Numerical simulations of a blood model under a confined linear shear flow reveal that confinement markedly modifies the properties of blood flow. A nontrivial spatiotemporal organization of blood elements is shown to trigger hitherto unrevealed flow properties regarding the viscosity η, namely ample oscillations of its normalized value ½η ¼ ðη − η 0 Þ=ðη 0 ϕÞ as a function of hematocrit ϕ (η 0 ¼ solvent viscosity). A scaling law for the viscosity as a function of hematocrit and confinement is proposed. This finding can contribute to the conception of new strategies to efficiently detect blood disorders, via in vitro diagnosis based on confined blood rheology. It also constitutes a contribution for a fundamental understanding of rheology of confined complex fluids. Introduction.-Blood flow in microcirculation is essential for delivery of nutrients and removal of metabolic waste products to or from tissues. These functions are ensured by proper regulation of blood flow down to the capillary level. One of the main factors controlling capillary circulation is microvascular resistance to blood flow. This effect, in spite of extensive investigation, is still to be fully elucidated, and some fundamental issues remain open. Blood is to good approximation a suspension of red blood cells (RBCs). Blood rheology is dictated by dynamics of RBCs and their interaction with blood vessel walls. A significant research effort has been devoted so far to macroscopic rheology [1,2]. Most of the research on rheology in confined geometries has focused on the famous Fahraeus-Lindqvist (FL) effect [3][4][5] (see recent review [6]), where confinement has been shown to strongly affect the rheology, with a decrease in apparent viscosity as the diameter of a vessel decreases. These advances have not exhausted yet the intricate behavior inherent to rheology of confined blood, as reported in this Letter.A property that is commonly of interest for nonconfined suspensions is the viscosity as a function of the volume fraction ϕ, ηðϕÞ. In the dilute regime, i.e., when hydrodynamic interactions between suspended entities can be neglected, η takes the generic form η ¼ η 0 ð1 þ a 1 ϕÞ, where η 0 is the viscosity of the suspending fluid and a 1 is a quantity (the so-called intrinsic viscosity), that depends, in general, on the properties of the suspension. For example, for rigid particles, a 1 is just a universal number and is equal to 5=2; this is the famous Einstein result [7,8]. a 1 was calculated by Taylor [9] for emulsions, and extended to vesicle suspensions (a blood model) quite recently [10]. When the volume fraction increases, hydrodynamic interactions among suspended