Previous research has shown that behavioural mimicry fosters affiliation, and can be used to infer whether people belong to the same social unit. However, we still know very little about the generalizability of these findings and the individual factors involved. The present study intends to disentangle two important variables and assess their importance for affiliation: the matching in time of the behaviours versus their matching in form. In order to address this issue, we presented participants with short videos in which two actors displayed a set of small movements (e.g. crossing their legs, folding their arms, tapping their fingers) arranged to be either contingent in time or in form. A dark filter was used to eliminate ostensive group marks, such us phenotype or clothing. Participants attributed the highest degree of affiliation to the actors when their subsequent movements matched in form, but were delayed by 4-5 seconds, and the lowest degree when the timing of their movements matched, but they differed in form. To assess the generalizability of our findings, we took our study outside the usual Western context and tested a matching sample of participants from a traditional small-scale society in Kenya. In all, our results suggest that movements are used to judge the degree of affiliation between two individuals in both large- and small-scale societies. While moving in different ways at the same time seems to increase the perceived distance between two individuals, movements which match in form seem to invoke closeness.