Cracked tooth syndrome (CTS) is a common clinical finding for teeth, it affects about 5% of all adults each year. The finding of CTS is favored by several risk factors such as restorations, bruxism, occlusion habits, and age. Treatment options range, depending on the severity, from no treatment at all to tooth extraction. Early diagnosis of CTS is crucial for optimal treatment and symptom reduction. There is no standard procedure for an evidence-based diagnosis up to date. The diagnosis is a challenge by the fact that the symptoms, including pain and sensitivity to temperature stimuli, cannot be clearly linked to the disease. Commonly used visual inspection does not provide in-depth information and is limited by the resolution of human eyes. This can be overcome by magnifying optics or contrast enhancers, but the diagnosis will still strongly rely on the practicians experience. Other methods are symptom reproduction with percussions, thermal pulp tests or bite tests. Dental X-ray radiography, as well as computed tomography, rarely detect cracks as they are limited in resolution. Here, we investigate X-ray dark-field tomography (XDT) for the detection of tooth microcracks. XDT simultaneously detects X-ray small-angle scattering (SAXS) in addition to the attenuation, whereas it is most sensitive to the micrometer regime. Since SAXS originates from gradients in electron density, the signal is sensitive to the sample morphology. Microcracks create manifold interfaces which lead to a strong signal. Therefore, it is possible to detect structural changes originating from subpixel-sized structures without directly resolving them. Together with complementary attenuation information, which visualizes comparatively large cracks, cracks are detected on all length-scales for a whole tooth in a non-destructive way. Hence, this proof-of principle study on three ex-vivo teeth shows the potential of X-ray scattering for evidence-based detection of cracked teeth.