Although some studies in tropical forests suggest the existence of elevational discontinuities, zones or critical altitudes in floristic composition (briefly reviewed by Lovett 1996) two data sets from the Usambara mountains of Tanzania independently show a continuous elevational change in moist forest large tree communities (Hamilton et al. 1989, Lovett 1996). However, one data set is from the West Usambara mountains (Lovett 1996) and the other is from the East Usambara (Hamilton et al. 1989). The junction of the two data sets at around 1000–1200 m in elevation is regarded as a critical altitude (Lovett 1996), and so neither conclusively prove continuity over the entire elevational range of the forests. The two data sets were collected using the same methodology, the twenty-tree variable-area plotless technique (Hall 1991), and together constitute a transect covering nearly 2 km in elevation. They are combined here to test for continuity in change in forest large-tree communities from lowland to upper montane forest types.