This study compared the effect of postexercise water immersion (WI) at different temperatures on common femoral artery blood flow (CFA), muscle (total haemoglobin; tHb) and skin perfusion (cutaneous vascular conductance; CVC), assessed by Doppler ultrasound, near-infrared spectroscopy (NIRS) and laser Doppler flowmetry, respectively. Given that heat stress may influence the vascular response during cooling, nine men cycled for 25 min at the first ventilatory threshold followed by intermittent 30-s cycling at 90% peak power until exhaustion at 32·8 ± 0·4°C and 32 ± 5% RH. They then received 5-min WI at 8·6 ± 0·2°C (WI9 ), 14·6 ± 0·3°C (WI15 ), 35·0 ± 0·4°C (WI35 ) or passive rest (CON) in a randomized, crossover manner. Heart rate (HR), mean arterial pressure (MAP), muscle (Tmu ), thigh skin (Tthigh ), rectal (Tre ) and mean body (Tbody ) temperatures were assessed. At 60 min postimmersion, decreases in Tre after WI35 (-0·6 ± 0·3°C) and CON (-0·6 ± 0·3°C) were different from WI15 (-1·0 ± 0·3°C; P<0·05), but not from WI9 (-1·0 ± 0·3°C; P = 0·074-0·092). WI9 and WI15 had reduced Tbody , Tthigh and Tmu compared with WI35 and CON (P <0·05). CFA, tHb and CVC were lower in WI9 and WI15 compared with CON (P<0·05). tHb following WI9 remained lower than CON (P = 0·044) at 30 min postimmersion. CVC correlated with tHb during non-cooling (WI35 and CON) (r2  = 0·532; P<0·001) and cooling recovery (WI9 and WI15 ) (r2  = 0·19; P = 0·035). WI9 resulted in prolonged reduction in muscle perfusion. This suggests that CWI below 10°C should not be used for short-term (i.e. <60 min) recovery after exercise.