Plasma and urine beta-thromboglobulin (BTG) were measured in 52 patients with established deep vein thrombosis (DVT) and in 100 patients with clinically suspected DVT but with a negative venogram. Both plasma BTG (geometric mean 54: 95% range 12--239 ng/ml) and urine BTG (0.25; 0.03--3.1 ng/ml) were significantly elevated (p less than 0.005) in patients with DVT compared to symptomatic patients with a negative venogram (plasma BTG 32, 9--112 ng/ml; urine BTG 0.12, 0.02-- 0.58 ng/ml). Sensitivity (35%) and specificity (80%) of the plasma BTG assay for the diagnosis of DVT were low. The urine BTG assay had a sensitivity of 37% but a specificity of 100%. There was a significant correlation between plasma and urine BTG (r = 0.68, p less than 0.005). Serial BTG measurements were made in urine (40 patients) and plasma (20 patients) from high-risk neurosurgical cases who were screened with 125I-fibrinogen leg scanning and impedance plethysmography. BTG was elevated postoperatively and returned to normal within 2 or 3 days, but rose again in 10 patients in association with the development of DVT. The rise of BTG preceded the uptake of 125I-fibrinogen and lasted for only a few days. The return to normal of BTG was not related to treatment with anticoagulants. While measurement of BTG in plasma and urine is of limited value in the clinical diagnosis of venous thrombosis, the data indicate platelet activation occurs in venous thrombosis, but is maximal or perhaps limited to the initial phase of thrombus development.