Undetected intravenous placement of epidural catheters is rare but potentially fatal and no perfect identification method exists. Epidural catheters may be flushed before insertion to identify faulty epidural catheters, or to prime the system with local anaesthetic. We hypothesised that flushing epidural catheters before insertion may delay the detection of intravenous placement. We investigated our theory using both in vitro and in vivo models. The in vitro component examined flowrates in flushed and unflushed epidural catheters, using conditions designed to mimic epidural venous pressure. The in vivo component examined the flow within flushed and unflushed epidural catheters inserted into the forearm veins of 20 anaesthetised patients, using a randomised crossover design. The end-point utilised for both components was the time taken for frank blood to reach the 20 cm mark on the epidural catheter. Blood flow to the 20 cm mark on the epidural catheter was significantly faster in the unflushed catheters than the flushed catheters, both in vitro and in vivo (in vitro, unflushed median = 18.6 s (range: 18.0 to 20.5 s), flushed 37.6 s (32.6 to 91.2 s), P=0.0009; in vivo, unflushed 9.2 seconds (range 5.0 to 35.3 s), flushed 19.2 s (10.6 to 47.4 s), P=0.003 in vivo). Flushed catheters also demonstrated a greater variability in the range of flowrates. Flushing epidural catheters before insertion may delay the detection of intravenous placement.