This paper is based on empirical research on a taxonomy of technological environmental innovations. It draws on a databank with over 500 examples of new technologies (materials, products, processes and practices) which come with benign environmental effects. The approaches applied to interpreting the datasets are innovation life cycle analysis, and product chain analysis. Main results include the following: 1. Innovations merely aimed at eco-efficienc y do in most cases not represent significant contributions to improving the properties of the industrial metabolism. This can better be achieved by technologies that fulfill the criteria of eco-consistency (metabolic consistency), also called eco-effectiveness. 2. Ecological pressure of a technology is basically determined by its conceptual make-up and design. Most promising thus are technologies in earlier rather than later stages of their life cycle (i.e. during R&D and customisation in growing numbers), because it is during the stages before reaching the inflection point and maturity in a learning curve where technological environmental innovations can best contribute to improving ecological consistency of the industrial metabolism while at the same time delivering their maximum increase in efficiency as well. 3. Moreover, environmental action needs to focus on early steps in the vertical manufacturing chain rather than on those in the end. Most of the ecological pressure of a technology is no rmally not caused end-of-chain in use or consumption, but in the more basic steps of the manufacturing chain (with the exception of products the use of which consumes energy, e.g. vehicles, appliances). There are conclusions to be drawn for refocusing attention from downstream to upstream in life cycles and product chains, and for a shift of emphasis in environmental policy from regulation to innovation. Ambitious environmental standards, though, continue to be an important regulative precondition of ecologically benign technological innovation.