The rate coefficient (k1) of the reaction between hydroxyl radical and hydroxyacetone, which remained so far controversial, was determined over the temperature range 290-500 K using pulsed-laser photolysis coupled to pulsed-laser induced fluorescence (PLP-PLIF). Hydroxyl radical was generated by pulsed photolysis of H2O2 at 248 nm. The results show that at a pressure of 50 Torr He, the rate coefficient obeys a negative temperature dependence k1(T) = (1.77 ± 0.19) × 10(-12) exp((353 ± 36)/T) cm(3) molecule(-1) s(-1) for temperatures between 290 and 380 K, in good agreement with the results of Dillon et al. (Phys. Chem. Chem. Phys. 2006, 8, 236) at 60 Torr He. However, always at 50 Torr He but for the higher temperature range 410-500 K, a positive temperature dependence was found: k1(T) = (1.14 ± 0.25) × 10(-11) exp(-(378 ± 102)/T) cm(3) molecule(-1) s(-1), close to the expression obtained by Baasandorj et al. (J. Phys. Chem. A 2009, 113, 10495) for pressures of 2 and 5 Torr He but at lower temperatures, 280-360 K, where their k1(T) values are well below these of Dillon et al. and of this work. Moreover, the rate coefficient k1(301 K) determined as a function of pressure, from 10 to 70 Torr He, shows a pronounced decrease once the pressure is below ∼40 Torr He, thus explaining the disparity between the higher-pressure data of Dillon et al. and the lower-pressure results of Baasandorj et al. The pressure dependence of k1 and of its temperature-dependence below ∼400 K is rationalized by the reaction proceeding via a hydrogen-bonded prereactive complex (PRC) and a submerged transition state, such that at high pressures collisionally thermalized PRCs contribute additional reactive flux over and through the submerged barrier. The high-pressure rate coefficient data both of Dillon et al. and of this work over the combined range 230-500 K can be represented by the theory-based expression k1(T) = 5.3 × 10(-20) × T(2.6) exp(1100/T) cm(3) molecule(-1) s(-1).