A common misconception among microsurgeons is that free tissue failure is an "all-or-none" phenomenon. In other words, there is an instantaneous cessation of blood flow to a flap primarily due to thrombosis at the arterial or venous anastomotic site (primary thrombosis) with complete flap loss as a result. Contrary to this belief, we have found that free tissue transfers occasionally "die" a slow, progressive, and partial death. This most likely is due to gradual "shutting down" of the microcirculation by the showering of microemboli downstream from the arterial anastomosis (secondary thrombosis). We discuss our clinical experience during the past 6 years with 10 patients in whom free flap failure was not an all-or-none phenomenon and describe the expectant management of these failing free flaps. Microvascular reconstruction of the lower extremities was performed for recalcitrant sickle cell ulceration, chronic venous stasis ulceration in a patient with anti-thrombin III deficiency, dry gangrene of the plantar surface of the foot and toes secondary to posterior tibial artery injection in an intravenous drug abuser, Gustillo type IIIc injury requiring reconstitution of the femoropopliteal artery after an automobile bumper injury, Gustillo type IIIb injury resulting from gunshot wounds, an open ankle joint in a patient with severe peripheral vascular disease, and osteomyelitis. Other cases included cheek soft tissue reconstruction after wide resection of a recurrent dermatofibrosarcoma protuberans of the parotid gland, chest wall coverage after claviculectomy for osteoradionecrosis, and thumb replantation complicated by refractory vasospasm. Latissimus dorsi (four patients), rectus abdominis (three patients), scapular (one patient), and radial forearm fasciocutaneous (one patient) flaps were used. Flap compromise was detected between 4 hours and 6 weeks after surgery. Salvage attempts included Fogarty thrombectomy (four patients), anastomotic revision (two patients), streptokinase instillation (one patient), and leech application (one patient). Failing free flaps were managed expectantly with the use of daily dressing changes. This allowed for the survival of sufficient soft tissue to effect coverage of exposed bone, tendon, or joint. Hyperbaric oxygen was not administered. Tangenital excision of the eschar was performed between 6 and 51 days after free tissue transfer. Skin grafting was delayed for 15 to 80 days. In all cases, a successful outcome was achieved ultimately by either a single skin-grafting procedure (seven patients) or groin flap coverage (one patient). Follow-up ranged from 2 to 35 months.(ABSTRACT TRUNCATED AT 250 WORDS)