Variation in the effects of disturbance across three differently aged glacial terraces in a terrace rimu (Dacrydium cupressinum) forest were investigated. Disturbance histories were reconstructed and associated patterns of stand development, differences in stand composition and stand structure determined. This forest variation was related to environmental variation between the differently aged terraces. Forest stand composition was investigated across the three terraces using Indicator Species Analysis and Detrended Correspondence Analysis for data collected from ninety stands. This data was used to categorise forest community composition. Analysis of stand structure focused on the dominant stand type on each of the three terraces for which disturbance histories and patterns of stand development were determined. Disturbance histories were reconstructed form size and age-class frequency distributions, and the spatial distribution of tree ages to identify even-aged groups of trees that had established in response to past disturbance. Soil drainage and stage of soil development were determined for the dominant stands. Differences in stand structure between the three main stands reflected differences in disturbance history which were associated with different soil profile and soil drainage characteristics. For stand S1 on the oldest terrace the spatial distribution of tree ages showed that catastrophic disturbance had occurred on the most poorly drained and deepest soils. For stand S2 on the middle aged terrace sporadic gap formation had occurred on less poorly drained and shallower soils. For stand S3 on the youngest terrace progressive canopy mortality had occurred on the shallowest soils, of a similar drainage to those in stand S2. Variation in the soil characteristics between terraces reflected a secular trend in soil development towards increased depth and poorer drainage. Stand composition was primarily influenced by disturbance history. The effects of different disturbance histories on stand composition was inferred from species regeneration patterns, the distribution of seedlings beneath different canopy cover classes, and gap regeneration characteristics. The least shade tolerant species, Dacrydium cupressinum, Lagarostrobos colensoi and Phyllocladus alpinus favoured regeneration after infrequent catastrophic disturbance. While the more shade tolerant Prumnopitys ferruginea, Weinmannia racemosa and Quintinia acutifolia favoured establishment following more frequent canopy gap formation. Differences in stand composition also reflected differences in soil drainage, interspecific competition, and stand structure. The association of species with differently drained soils within stands and between stands suggested Lagarostrobos colensoi and Phyllocladus alpinus favoured the most poorly drained soils, while Prumnopitys ferruginea favoured less poorly drained soils. Interspecific competition accounted for ongoing compositional changes where Dacrydium cupressinum appeared to be competing out Phyllocladus alpinus and Lagarostrobos colensoi on account of its greater stature and longevity. Stand structure affected species richness. The least number of species was associated with the most uniform stands while the highest number of species was associated with the least uniform stands. Uniform stands were presumed to reduce the chance of colonisation by differentially adapted species. Variation in the effects of disturbance explained changes in stand structure and composition across the terraces. A secular change in the stand dynamics of terrace rimu forest is envisaged over millennia reflecting soil development, where changes in forest composition and stand structure reflect the interaction of disturbance and soil drainage.