We measured the concentrations of extractable bioactive compounds in heartwood of live yellow-cedar (Chamaecyparis nootkatensis) trees and five classes of standing snags (1-5, averaging 4, 14, 26, 51, and 81 years-since-death, respectively) to determine how the concentrations changed in the slowly deteriorating snags. Three individuals from each of these six condition classes were sampled at four sites spanning a 260-km distance across southeast Alaska, and the influence of geographic location on heartwood chemistry was evaluated. Cores of heartwood were collected at breast height and cut into consecutive 5-cm segments starting at the pith. Each segment was extracted with ethyl acetate and analyzed by gas chromatography. Concentrations of carvacrol, nootkatene, nootkatol, nootkatone, nootkatin, and total extractives (a sum of 16 compounds) for the inner (0-5 cm from pith), middle (5-10 cm from pith), and surface (outer 1.1-6.0 cm of heartwood) segments from each core were compared within each tree condition class and within segments across condition classes. Heartwood of class 1 and 2 snags had the same chemical composition as live trees. The first concentration changes begin to appear in class 3 snags, which coincides with greater heartwood exposure to the external environment as decaying sapwood sloughs away, after losing the protective outer bark. Within core segments, the concentrations of all compounds, except nootkatene, decrease between snag classes 2 and 5, resulting in the heartwood of class 5 snags having the lowest quantities of bioactive compounds, although not different from the amounts in class 4 snags. This decline in chemical defense is consistent with heartwood of class 5 snags being less decay-resistant than heartwood of live trees, as observed by others. The unique heartwood chemistry of yellow cedar and the slow way it is altered after death allow dead trees to remain standing for up to a century with a profound impact on the ecology of forests in southeast Alaska where these trees are in decline.