The present study was conducted in subtropical humid forests of Meghalaya to study the distributional pattern of species, floristic composition, community structure and tree population structure. Forest fragments of varying sizes (0.5 ha, 1 ha, 2 ha and 5 ha) were used in the study. All of the forest fragments are distributed within the same altitudinal range, and had similar rainfall and temperature regimes. Four forest fragments were sampled using random quadrats to analyze the impact of fragment size on tree diversity and population structure. Indices were used to compare the dispersion pattern of plant species, species diversity among fragments, and the heterogeneity and homogeneity of the fragments. A total of 45 tree species were recorded from all the fragments and simple correlation showed that the species richness was positively related to fragment size (n = 4, P<0.05). Other measures such as stand density, basal cover, diversity and dominance indices were unrelated to fragment size. Forest canopy was composed mainly of Castonopsis indica, C. armata, Schima khasiana and Engelhardtia spicata in all of the fragments. A majority of woody tree species were contagiously distributed which made the forest community highly patchy in nature. Low values of Sorensen’s similarity index and high values of Whittaker’s β-diversity index between the fragments suggest marked difference in the species composition among the fragments. The maximum stand density, basal area, Shannon diversity index and Simpson dominance index was recorded in a 5-ha fragment. Trees in lower girth classes (15–55 cm circumference at breast height, CBH) accounted for 52–86% of stand density in all fragments except in 0.5-ha fragments where middle girth class (55–95 cm CBH) accounted for 47% of the total stand density. The overall density of seedlings, saplings and adult trees formed a pyramidal structure, indicating that the period between the sapling and adult stage was a critical stage in tree life cycles, as the maximum mortality occurred during this period. These groves are mildly/highly disturbed by communities using forests for their timber and fuel wood requirements as well as cattle and goat grazing. Tree regeneration was apparently not limited, as was evidenced by high density of saplings. These human activities may have been responsible for a decrease in tree diversity in the smaller fragments. Large fragments that were less disturbed harbored species not present in small fragments.