This work investigates the impact of human-human interaction on a target behaviour change by comparing the effect of programme delivery of a fixed set of tailored information via proximate but trivial interaction between subjects and volunteers, a set of well-produced single-topic colour leaflets, and a control on behaviour in a real world setting. The behaviour targeted was householder sorting of food waste into specialised bins in high-rise apartment buildings in Shanghai, China, measured by discreet direct daily weighing of waste fractions. The unit of analysis was the set of households in each building. Two versions of the volunteer delivery were trialled: one neutral in tone and action, and the second slightly positive in tone and action. Despite the existence of tens of theories about behaviour change and hundreds of empirical case studies of pro-environmental behaviour change programs, human-human interaction is not mentioned as a predictor and is only rarely as possibly moderating subsequent conduct. Results suggest that human-human interaction is not likely to be a key explanatory factor but that a positive human-human interaction may be an important factor. Furthermore, these effects were observed when the six social influence mechanisms suggested in current behaviour theory were eliminated, suggesting new mechanisms need to be proposed. For practitioner waste managers the results indicate that funding programmes with human interaction may not be sufficient for greater results: the humans may need training for positive interaction. In addition, the results indicate that currently held opinions by theorists and practitioners on the relative usefulness of tailored information may need revising, since most compound it with human interaction. Explanatory studies are thus called for at programme and individual levels.