This paper examines the performance of rural groups in Kenya and addresses the methodological issues and challenges faced in doing this, and presents the empirical evidence regarding various hypothesized explanatory factors for relative performance levels. Eighty-seven groups and 442 households were surveyed using several approaches. Various performance measures were tested. Both descriptive analysis and regression models were used to gain a better understanding of the group-level and household-level factors that explain performance. Collective action is desired and practiced for a large number of tasks. The findings highlight the incredible number, diversity and dynamic nature of groups in the highlands of Kenya (and we suspect this finding is not terribly unique to this region). Assessing and comparing performance across a range of group activities is wrought with difficulties related to measurement and standardization. Focusing on groups undertaking similar activities makes it easier to delve more deeply into performance drivers. The empirical analysis focused on the effect of group structural variables (e.g. its size) on performance. We found that choice of performance measure and level at which it is measured (e.g. household, group) matters when it comes to trying to explain the variability in that measure. An analysis across different types of groups engaged in exactly the same activity (tree nurseries) found that predicted group performance was not linked to any easy-to-measure group characteristic, implying that for this task dissemination need not be targeted towards particular types of groups. Looking more broadly at a range of activities, we found that structural factors had varied results.