The vast majority of microfluidic devices are developed in PDMS by molding ("soft lithography") because PDMS is an inexpensive material, has physicochemical properties that are well suited for biomedical and physical sciences applications, and design cycle lengths are generally adequate for prototype development. However, PDMS molding is tediously slow and thus cannot provide the high- or medium-volume production required for the commercialization of devices. While high-throughput plastic molding techniques (e.g. injection molding) exist, the exorbitant cost of the molds and/or the equipment can be a serious obstacle for device commercialization, especially for small startups. High-volume production is not required to reach niche markets such as clinical trials, biomedical research supplies, customized research equipment, and classroom projects. Crucially, both PDMS and plastic molding are layer-by-layer techniques where each layer is produced as a result of physicochemical processes not specified in the initial photomask(s) and where the final device requires assembly by bonding, all resulting in a cost that is very hard to predict at the start of the project. By contrast, stereolithography (SL) is an automated fabrication technique that allows for the production of quasi-arbitrary 3D shapes in a single polymeric material at medium-volume throughputs (ranging from a single part to hundreds of parts). Importantly, SL devices can be designed between several groups using CAD tools, conveniently ordered by mail, and their cost precisely predicted via a web interface. Here we evaluate the resolution of an SL mail-order service and the main causes of resolution loss; the optical clarity of the devices and how to address the lack of clarity for imaging in the channels; and the future role that SL could play in the commercialization of microfluidic devices.