Purpose Clean cookstoves have emerged over the past half century as an important technological innovation to reduce indoor air pollution from cooking with traditional fuels. However, widespread adoption remains elusive, suggesting the need for other measures to accompany dissemination of clean cookstoves. Despite knowledge about health impacts of cookstove smoke and a body of evidence pointing to the efficacy of health education for supporting behaviour change, health messaging is relatively unexplored in the cookstove sector. This paper aims to present findings from action research in Cambodia that investigates how social innovation around positive and negative health messaging influences demand for clean biomass cookstoves. Design/methodology/approach An action research approach was taken, involving the design and implementation of a health marketing campaign alongside promotion of a clean burning biomass cookstove. Four communes were assigned as intervention communes and a fifth as the control group. Among the four intervention communes, two were provided with positive health messaging and the other two with negative health messaging. The methods included a baseline study of 381 households using structured surveys, roll-out of the health campaign, in-depth interviews with households and sales agents, ten focus group discussions with households and an endline structured survey of all 381 households. Findings Neither the type (positive/negative) nor the intensity of the health campaigns had a significant impact on stove sales. Sales results show no pattern in either variable, and sales in the control commune were not lower than in communes where health campaigns were used. However, health messaging did increase awareness about health impacts of cooking with traditional biomass burning stoves. For almost all communes, in particular those that received positive-tone messages, an increased awareness of the health impact of cooking with traditional biomass burning stoves was observed. Cookstove price and personal characteristics of individual sales agents were shown to be the strongest factors affecting sales. Research limitations/implications The study relied on sales agents to deliver pre-assigned health messages. However, some sales agents did not follow instructions in delivering the messages, which made it difficult to compare the efficacy of the different campaign approaches. Due to a delay in the study, the campaign overlapped with the planting season when disposable incomes of famers is typically limited, reducing their ability to purchase a new cookstove. The 10-week duration of the campaign may not have been long enough to see an impact on sales, particularly for a product viewed as expensive for the average consumer. Practical implications The skill and motivation of individual sales agents can greatly affect cookstove marketing campaigns. The efficacy of individual sales agents appears to have been the strongest factor affecting sales, with the most successful sales agent using a combination of messages, including health information, to convince households to purchase the stoves. This warrants further study; designers of stove promotion campaigns might be able to learn from the behaviour and strategies of highly effective sales agents. Price continues to be an important factor influencing the adoption of clean cookstoves. In Cambodia, the main drivers of cookstove purchase (beside the sales agent) were availability of disposable income, time and fuel saved. Social implications Health messaging was shown to have minimal effect on cookstove purchase. Indeed, it is clear that cookstove adoption is influenced by multiple factors linked in complex ways. This is a very important finding for public health workers, who need to think more broadly about how they achieve the public health goals associated with cleaner cooking through approaches that do not necessarily focus on individual health goals. Originality/value Although health messaging has shown promise as an approach for supporting behaviour change in other sectors, it has not been widely studies in the context of clean cookstove adoption. This paper contributes to filling this gap and suggests some lines of enquiry for future research. The study pioneered innovative methods such as action research, use of graphic images and using established local sales agents as means of communicating messages about the health risks of cooking smoke and the benefits of improved cookstoves.