Plastics of sufficient purity reflect infrared light in very characteristic patterns. In contrast, other materials commonly found in municipal rubbish (such as paper, ceramics and glass) do not exhibit such patterns. Hence, a computer can be programmed to identify plastic items in a stream of rubbish based on data it receives from an infrared camera. Nor can identification be affected by possible obscuring of the infrared beam by large objects, or by the effects of moisture. The software developed compensates for such phenomena. This automated process should permit local councils to recycle a much greater volume of plastic than they currently do, without resorting to time-consuming manual sorting, either in the home, or at the waste disposal unit. This in turn, should lead to a general reduction in the consumption of finite oil reserves by the plastics industry.