We analyzed the correlation between thrombus regression on control venography performed after discontinuation of heparin therapy and recurrent venous thromboembolism (VTE) detected during clinical follow‐up in randomized trials comparing low‐molecular‐weight heparin (LMWH) and unfractionated heparin (UFH) in patients with deep vein thrombosis (DVT). Data were ed from MEDLINE, conference s and reference lists of previous reviews. Randomized, controlled trials comparing LMWH and UFH for the treatment of DVT using a combined venographic and clinical assessment and with at least 2 months of follow‐up were selected. The proportions of patients with thrombus regression on control venography performed soon after discontinuation of heparin therapy and recurrent VTE at 2–6 months were independently collected by two researchers. Thirteen studies met the inclusion criteria. There was a strong inverse correlation between thrombus regression and recurrent VTE (r = − 0.70; P = 0.008). The venographic effect varied between the different LMWHs (P = 0.013). A very strong correlation was found when the results were pooled by the type of LMWH used (r = − 0.84; P = 0.037). No influence of the dose interval used on the venographic effect (P = 0.156) or on recurrent VTE (P = 0.218) was shown. The lack of thrombus regression in venography, performed soon after heparin discontinuation, was correlated with clinical recurrence. Non‐invasive imaging techniques should be relevant to identify non‐responders and to assess the optimal duration of initial heparin treatment in daily clinical practice.