The potential energy surface of the radical-molecule reaction C 2 H 3 + H 2 O in the gas phase is explored at the 6-31G(d,p) and 6-311G(d,p) B3LYP and single-point QCISD(T)/6-311G(2df,p) levels. The most favorable channel is the direct H-abstraction from H 2 O to C 2 H 3 leading to product P 1 C 2 H 4 + OH, whereas the other channels leading to the products P 2 CH 3 + CH 2 O, P 3 CH 3 CHO + H, P 4 cis-CH 2 CHOH + H and P 4 0 trans-CH 2 CHOH + H are kinetically much less competitive. For the direct H-abstraction channel, high-level energetic calculations at the QCISD(T)/6-311G(2df,p), QCISD(T)/6-311+G(2df,2p) and G2 levels using the B3LYP/6-31G(d,p) and QCISD/6-31G(d,p) optimized geometries are further performed to estimate the thermal rate constants over a wide temperature range 200-5000 K for comparison with future laboratory measurements. The calculated barrier heights at the QCISD(T)/6-311+G(2df,2p) and G2 levels based on the QCISD/6-31G(d,p) geometries with zero-point vibrational energy (ZPVE) correction are 12.6 and 13.0 kcal mol À1 , respectively. The results indicate that the C 2 H 3 + H 2 O reaction might play an important role at high temperatures (T > 1800 K) in the presence of gaseous water and should be incorporated in the C 2 H 3 -modeling of hydrocarbon-fuel combustion processes. Discussions are also made in comparison with the analogous reactions C 2 H 3 + H 2 and C 2 H + H 2 O. While the addition-elimination mechanism of another important radicalmolecule reaction C 2 H 4 + OH has been the subject of extensive theoretical and experimental studies, its Habstraction process leading to C 2 H 3 + H 2 O has received little attention. For the C 2 H 4 + OH ! C 2 H 3 + H 2 O channel, our calculations predict ZPVE-corrected barriers, 5.6 and 5.4 kcal mol À1 , respectively, at the QCISD(T)/6-311+G(2df,2p)//QCISD/6-31G(d,p) and G2//QCISD/6-31G(d,p) levels, and reveal its importance at high temperatures (T > 560 K). In the range 720-1173 K, the calculated high-level rate constants are quantitatively in good agreement with the measured values. However, our calculated activation energy, 9.5 and 9.3 kcal mol À1 at the QCISD(T)/6-311+G(2df,2p)//QCISD/6-31G(d,p) and G2//QCISD/6-31G(d,p) levels with ZPVE correction, respectively, suggests that the experimentally determined value 4-5 kcal mol À1 may be underestimated and future rate constant measurements over a wide temperature range including T > 1200 K may be desirable.