The origin of micro-diamonds is controversial and although the application to determine the grade and value of macro-diamonds in kimberlite/lamproite bodies continues to receive widespread usage there are several outstanding factors generally not considered, the most important of which is genesis. The issue is addressed in this study in the context that two classes of small diamonds (generally <0.5 mm and rarely <1 mm) are recognized. Micro-diamonds sensu-stricto (MDS) are typically sharp-edged octahedra, free of mineral inclusions and surface etching or corrosion, increase exponentially with decreasing size and are in overwhelming larger concentrations, by orders of magnitude, relative to macro-diamonds (>0.5 mm). The second class of small diamonds (SD <0.5 mm), used in industrial applications, may have modified solution-growth morphologies (e.g. dodecahedra, tetrahexahedra and related forms), and include loosely bonded polycrystalline diamonds (framesite), boart, fibrous cubes and broken fragments. There are large differences in volume to surface-area ratios between MDS and SD, demonstrating unequivocally that pristine and solution-modified forms could not have co-existed in equilibrium under the same P-T-t-fO2 conditions in the mantle. From detailed studies of N and C in diamond, and experimental results on the redox-partitioning of N in the presence of metallic Fe, it is concluded that MDS are plume-related from the D″ core-mantle boundary, and are melt-derived in lower mantle proto-kimberlite. The lower mantle is expectedly saturated in metallic Fe, and is highly depleted in N which is siderophile under very low f O2 conditions, a setting in which excessively large (∼100 to 3000 ct), but rare Type II mega-diamonds (but also MDS) are inferred to have originated. These diamonds (Type II, Ib, IaA) are distinct from the majority of N-rich Type Ia upper mantle macro-diamonds that grew slowly by metasomatic processes and annealed over long periods. Two crystal growth laws are possibly applicable to the size-distribution of diamonds encountered in kimberlites/lamproites. Gibrat’s Law of proportionate, short-term crystal growth in open systems by advection is applicable to magmatic MDS, whereas macro-diamonds bear some relation to McCabe’s Law of long-term, relatively constant crystal growth, by diffusion metasomatism. The range from small to large diamonds (SFD size-frequency-distribution) is lognormal but is composed of two segments: the smaller size (<0.5 mm) fraction has an overall linear distribution, whereas macro-diamonds (>0.5 mm) are quadratic. The two distributions meet or overlap in a marked discontinuity, implying but not proving distinct origins. The power law governing SFD lognormal distributions is fundamental and is widespread across an enormous number of disciplines (from biology to economics), and may be universal (e.g. it is applicable to planetary scale meteorite impact craters, and to the SFD of cosmic-diamonds from supernovae explosions). Industry applications in resource predictions are from mixtures of diamonds (MDS and SD), and extrapolation to larger stones is valid because the fundamental law is independent of origins.