Fibre sourcing is a critical strategic question for all sawmills and pulpmills, but the degree of supply integration though long-term contracts and forest ownership varies widely. The purpose of this research was to investigate the extent to which forest-mill integration patterns can be explained by the transaction cost economics (TCE) theory. TCE theory holds that organizations will choose transaction governance forms that minimize transaction costs. The TCE factors expected to influence that choice can be grouped into three categories; transaction frequency, market uncertainty, and asset specificity. Interviews with various industry representatives suggested that factors from all three categories are relevant to the question of forest-mill integration. A survey was conducted of mills in New Zealand and Sweden, providing data on their supply mix and various TCE factors. Of an estimated population of approximately 450 mills, 136 mills were sampled and 88 responded to the survey. Fractional logit models were developed to explore the factors that may influence the integration decision. Considerable evidence was found for the importance of TCE factors in driving fibre supply integration. The evidence was strongest for factors related to asset specificity, including forest owner concentration and the specificity of a mill's fibre requirements. Transaction frequency appears less important; while integration was found to be significantly associated with the number of mills an organisation has within the supply basin, the influence of mill capacity was found to vary. There was weak evidence for the importance of uncertainty, and perhaps only through the impact of forest owner concentration on market conduct. Integration was found significantly higher for pulpmills than sawmills, and higher in Sweden than in New Zealand. The latter result is difficult to explain by TCE theory, and suggests that non-TCE factors play a significant role. Survey responses also indicated that non-TCE factors are important. Further research is required to enlarge the sample size and better understand the role of TCE factors in forest-mill integration.