It is often assumed that dominant forest understory communities are predictably associated with overstory tree species, yet several long-term studies suggest that understory communities are more independent of overstory change. We use a 10-year dataset to explore variation in understory communities in a mature second-growth temperate forest in western Washington. We classify all recorded species into six growth-forms (graminoids, ferns, shrubs, subshrubs, saplings, herbaceous species), and introduced species (collectively grouped) to analyze responses to overstory productivity, stand age, canopy heterogeneity, soil type, stand type, and proximity to a canopy-gap forming pathogen (Phellinus sulphurascens), as well as overstory carbon (C) changes through the decade. Plant diversity and percent cover declined marginally through time (Shannon's H′ declined by 8% and percent cover by 4.5%), as plots remained dominated by the species Polystichum munitum and Gaultheria shallon. Species richness decreased significantly by 23% among years (mean plot richness was 10.02 in 2008 compared to 8.05 in 2018), and diversity, subshrubs, and shrubs generally declined with stand age. Shrubs were more abundant in conifer-dominated plots. Ferns, and changes in ferns, were associated with presence of P. sulphurascens, where cover increased in infected plots. Ordination results suggested community composition was correlated with changes in canopy cover (conifer forests) and stand age (deciduous forests). Changes in total plot C and canopy cover were also associated with diversity and total percent cover. Nevertheless, our results support incremental changes in understory communities on decadal time scales and limited predictability of the understory based on the overstory.