Context. Material processed by the CNO cycle in stellar interiors is enriched in 17O. When mixing processes from the stellar surface reach these layers, as occurs when stars become red giants and undergo the first dredge up, the abundance of 17O increases. Such an occurrence explains the drop of the 16O/17O observed in RGB stars with mass larger than 1.5 M_solar. As a consequence, the interstellar medium is continuously polluted by the wind of evolved stars enriched in 17O . Aims. Recently, the Laboratory for Underground Nuclear Astrophysics (LUNA) collaboration released an improved rate of the 17O(p,alpha)14N reaction. In this paper we discuss the impact that the revised rate has on the 16O/17O ratio at the stellar surface and on 17O stellar yields. Methods. We computed stellar models of initial mass between 1 and 20 M_solar and compared the results obtained by adopting the revised rate of the 17O(p,alpha)14N to those obtained using previous rates. Results. The post-first dredge up 16O/17O ratios are about 20% larger than previously obtained. Negligible variations are found in the case of the second and the third dredge up. In spite of the larger 17O(p,alpha)14N rate, we confirm previous claims that an extra-mixing process on the red giant branch, commonly invoked to explain the low carbon isotopic ratio observed in bright low-mass giant stars, marginally affects the 16O/17O ratio. Possible effects on AGB extra-mixing episodes are also discussed. As a whole, a substantial reduction of 17O stellar yields is found. In particular, the net yield of stars with mass ranging between 2 and 20 M_solar is 15 to 40% smaller than previously estimated. Conclusions. The revision of the 17O(p,alpha)14N rate has a major impact on the interpretation of the 16O/17O observed in evolved giants, in stardust grains and on the 17O stellar yields.