We previously reported that low‐dose doxycycline (DOXY) therapy reduces host‐derived collagenase activity in gingival tissue of adult periodontitis (AP) patients. However, it was not clear whether this in vivo effect was direct or indirect. In the present study, inflamed human gingival tissue, obtained from AP patients during periodontal surgery, was extracted and the extracts partially purified by (NH4)₂SO₄ precipitation. The extracts were then analyzed for collagenase activity using SDS‐PAGE/fluorography/laser densitometry, and for gelatinase activity using type I gelatin zymography as well as a new quantitative assay using biotinylated type I gelatin as substrate. DOXY was added to the incubation mixture at a final concentration of 0–1000 μM. The concentration of DOXY required to inhibit 50% of the gingival tissue collagenase (IC₅₀) was found to be 16–18 μM in the presence or absence of 1.2 mM APMA (an optimal organomercurial activator of latent procollagenases); this IC₅₀ for DOXY was similar to that exhibited for collagenase or matrix metalloproteinase (MMP)‐8 from polymorphonuclear leukocytes (PMNs) and from gingival crevicular fluid (GCF) of AP patients. Of interest, Porphyromonas gingivalis collagenase was also inhibited by similar DOXY levels (IC₅₀= 15 μM), however the collagenase activity observed in the gingival tissue extracts was found to be of mammalian not bacterial origin based on the production of the specific α^A (3/4) and α^B (1/4) collagen degradation fragments. In contrast, the inhibition of collagenase purified from culture media of human gingival fibroblasts (MMP‐1) required much greater DOXY levels (IC₅₀=280 μM). The predominant molecular forms of gelatinolytic activity present in the AP patients gingival tissue extracts were found to closely correspond to the 92 kD PMN‐type gelatinase (MMP‐9) although small quantities of 72 kD fibroblast‐type gelatinase (MMP‐2), and some other low molecular weight gelatinases, were also detected. The IC₅₀ of DOXY versus gingival tissue gelatinolytic activity was estimated at 30–50 μM measured using either type I gelatin zymography or the biotinylated type I gelatin assay. We conclude that MMPs in inflamed gingival tissue of AP patients, like those in GCF, originate primarily from infiltrating PMNs rather than resident gingival cells (fibroblasts and epithelial cells) or monocyte/macrophages, and that their pathologically‐elevated tissue‐degrading activities can be directly inhibited by pharmacologic levels of doxycycline.