This paper consists of two parts - a detailed experimental investigation on the activity of supported threonine aldolase for use in a microreactor/flow reactor and a generic, explorative outline on potential productivity of enzymatic microreactors; with the need to use data from other enzymes here as well. Threonine aldolase was immobilized on different supports foreseen to be coated or packed within the micro- and millichannels of flow reactors. As enzyme supports, an innovative high-surface area and high-porosity silica (Nanosprings), the commercial Eupergit CM, and polymer membrane materials were used. For Eupergit CM, the share of chemically fixed enzymes and active enzymes were within the reported literature performance for enzyme immobilization. The values of immobilized enzymes on Nanosprings were somewhat lower. However, this is overcompensated by the larger specific surface area of Nanosprings and the availability of a new, second generation flow reactor concept, which involves a tight packing of stacked Nanosprings disks. For the investigated membranes, the density of immobilized enzymes was close to the reported value of capacity. From these experimental results, calculations on (maximally achievable) productivity of microreactors were made. This was combined with results of glucose oxidase immobilization on Nanosprings. The objective was to achieve high enzyme loadings to obtain reasonable space-time yields to ensure cost-effectiveness of the derived chemical reaction processes. These productivities were benchmarked to industrial production needs for two processes - a high-value pharmaceutical intermediate synthesis (towards chiral amino alcohols, using threonine aldolase) and a large-volume fine-chemical synthesis (the gluconic acid synthesis, using glucose oxidase), including suggestions for reactor scale-up. This shows that the present loading performance suffices in the case of high-value products, such as pharmaceuticals, but not for larger-volume, low-cost chemicals; at least under best-performance assumptions such as the avoidance of mass-transfer limitations. (C) 2012 Elsevier B.V. All rights reserved.