Deep vein thrombosis (DVT) is a common complication of spinal cord injury (SCI). Many specialized SCI nursing units use serial leg measurements to establish clinical diagnosis of DVT, in spite of extensive studies showing that clinical signs and symptoms of DVT have a specificity and sensitivity hardly better than chance. For 30 patients who spent 920 days in an SCI intensive care unit, we collected daily serial calf and thigh measurements and the results of radiofibrinogen uptake test (RFUT) performed about every third day. Three criteria for clinical diagnosis of DVT found in the literature were used: an increase from baseline, an increase from the previous day or a right-left difference of a specific minimal size. The cut-off points for minimal size used were also those suggested by the literature: over .5 inch, over 1 cm, or 1.2 cm and over for women, 1.5 cm and over for men. We found such changes or differences correlated weakly or not at all with RFUT results, and symptomatic increases or differences occurred very frequently. Major factors presumably underlying the variability of measurements and their lack of sensitivity and specificity are: premorbid leg asymmetry; atrophy of the legs after SCI; true changes in circumference due to factors other than DVT; and lack of measurement reliability. It was concluded taking serial leg measurements is of no value in the SCI population. For those reluctant to abandon tradition, suggestions are provided for improving the quality of measurements and related decision making. Further research is needed.