Abstract Arthropod community structure and herbivory were compared in replicate Douglas-fir and western hemlock canopies in intact old-growth (> 400-year-old), and Douglas-fir only in partially harvested old-growth, natural mature (150-year-old) stand, and regenerating plantations (10–20-year-old) in a 15 000 ha area including the H.J. Andrews Experimental Forest in western Oregon. Species diversity and abundances of several taxa, especially predators and detritivores, were significantly lower in plantations compared to older forests. Mature, old-growth, and partially harvested stands showed few significant differences, but principal components analysis suggested some differences in community structure and indicated that old-growth was least variable (tighter clustering) in arthropod diversity and abundance, whereas partially harvested stands were most variable. Defoliation was higher in the mature stands, probably because these stands were composed of relatively dense and pure Douglas-fir. Although old-growth appeared to be the source of greatest arthropod biodiversity in these forests, arthropod communities in Douglas-fir canopies may largely recover old-growth structure by 150 years, and partially harvested stands retain substantially greater arthropod diversity than do regenerating plantations.