Objective: To understand the impact of social organisation affiliation and farmers' agricultural production practices on farmer health. Organisations facilitate the acquisition and exchange of forms of social capital which can influence the adoption of practices with potential health impacts. In countries such as Ecuador, smallholder agriculture is practised by socially vulnerable populations. Agricultural production often involves the use of extremely hazardous pesticides, while practices that reduce the use of chemicals through integrated pest management (IPM) remain uncommon. Design: Longitudinal study (2007-2010). Setting: 12 Ecuadorian communities, previously part of a participatory action research study. Participants: 208 small-scale farmers. Inclusion criteria were: age between 18 and 65 years, literate and resident in the community for the previous 3 years. Primary outcomes: The differential effects of the membership in social organisations (as an effect modifier), on the relationship between the implementation of IPM practices (main independent variable) and farmers' health, measured by neurocognitive performance scores (better higher value; dependent variable). Results: Among organisational participants, the coefficient of association between the implementation of IPM practices for the category good/very good (vs no use) and neurocognitive performance, when farmers were involved in organisations, was negative and moderate (β=-0.17, SE 0.21) though not significant (p>0.1); for the category little/moderate use, the coefficient was positive (β=0.34, SE 0.19) and significant. Among those who did not participate in organisations, both little/moderate use and good/very good use of IPM practices were associated with an increase in neurocognitive performance. Conclusions: The effect of agricultural production practices on farmers' health, transmitted through organisations, can be differentiated. Organisations as structures of social capital seem to be functional in the social reproduction process of the communities studied. Results highlight the need to redirect the analysis of social capital to a more integrated study of social determination of health.