The work presented here demonstrates the extension of nuclear spin-noise-detected imaging  to 3D imaging. Spin noise detection provides a high potential for application at nano-sized samples, is implicitly free of rf-related artifacts and allows for fast recycling times owed to the independence from longitudinal magnetization recovery. Also no rf-related power restrictions need to be considered. Applying the enhancements and insights of over 10 years of spin noise research  the realization of this intrinsically insensitive experiment has become possible. The advancements include sliding window processing  of the continuously recorded time-domain data, avoiding T/R switching related ?transient effects? , and optimal tuning of the probe tuning for spin noise detection . With these improvements in the experimental setup and advanced data processing, it is now possible to record 3D nuclear spin-noise-detected images in a reasonable time frame using a state-of-the-art high resolution spectrometer. When recording conventional (rf-pulse excitation based) NMR experiments there is a trade-off between the resolution and the S/N ratio (which is determined mostly by the number of scans), which needs to be decided before starting the experiment. For spin-noise-detected experiments this trade-off can be adjusted as part of the data processing after the recording is complete. This involves novel iterative image reconstruction techniques based on the algebraic reconstruction algorithm . Even molecular imaging (spatially resolved spectroscopy) becomes feasible that way.