Forest management involves exploring through time the scheduling opportunities for timber and non-timber values within a forest. The ability to identify and form neighbourhoods - areas of suitable stand conditions and locations - is critical to this endeavour. This paper presents a GIS-based algorithm for identifying and forming forest neighbourhoods suitable for timber harvest scheduling. The resulting neighbourhoods are contiguous and overlapping, composed of stands sharing similar attributes. Similarity is based on a definable similarity list where stand conditions closest to one another in the list are most similar to one another. The algorithm is demonstrated with a single stand example and then a small forest example. Control of neighbourhood size is limited using a vector data model, except in forests composed of small stands. The examples illustrate that neighbourhood inclusion is dependent on both a forest stand's condition and relative position in the forest. The paper concludes with suggestions for further development of the algorithm.