Shifting cultivation in the Nghe An Province of Vietnam's Northern Mountain Region produces a characteristic land-cover pattern of small and larger fields. The pattern is the result of farmers cultivating either individually or in spatially clustered groups. Using spatially explicit agent-based modelling, and relying on empirical data from fieldwork and observations for parameterization of variables, the level of clustering in agricultural fields observed around a study village is reproduced. Agents in the model act to maximize labour productivity, which is based on potential yield and labour costs associated with fencing of fields, and are faced with physical constraints. The simulation results are compared with land-cover data obtained from remote sensing. Comparisons are made on patterns as detected visually and using the mean nearest-neighbour ratio. Baseline simulation outputs show high degrees of spatial clustering and similarity to the land-cover data, but also a need for further calibration of model variables and controls