The aim of this study was to investigate the effects of nocturnal railway noise on cardiovascular reactivity in young (25.8 +/- 2.6 years) and middle-aged (52.2 +/- 2.5 years) adults during sleep. Thirty-eight subjects slept three nights in the laboratory at 1-week interval. They were exposed to 48 randomized pass-bys of Freight, Passenger and Automotive trains either at an 8-h equivalent sound level of 40 dBA (Moderate) and 50 dBA (High) or at a silent Control night. Heart rate response (HRR), heart response amplitude (HRA), heart response latency (HRL) and finger pulse response (FPR), finger pulse amplitude (FPA) and finger pulse latency (FPL) were recorded to measure cardiovascular reactivity after each noise onset and for time-matched pseudo-noises in the control condition. Results show that Freight trains produced the highest cardiac response (increased HRR, HRA and HRL) compared to Passenger and Automotive. But the vascular response was similar whatever the type of train. Juniors exhibited an increased HRR and HRA as compared to seniors, but there was no age difference on vasoconstriction, except a shorter FPL in seniors. Noise level produced dose-dependent effects on all the cardiovascular indices. Sleep stage at noise occurrence was ineffective for cardiac response, but FPA was reduced when noise occurred during REM sleep. In conclusion, our study is in favor of an important impact of nocturnal railway noise on the cardiovascular system of sleeping subjects. In the limit of the samples studied, Freight trains are the most harmful, probably more because of their special length (duration) than because of their speed (rise time).