Context. Asymptotic giant branch (AGB) stars lose their envelopes by means of a stellar wind whose driving mechanism is not understood well. Characterizing the composition and thermal and dynamical structure of the outflow provides constraints that are essential for understanding AGB evolution, including the rate of mass loss and isotopic ratios. Aims. We characterize the CO emission from the wind of the low mass-loss rate oxygen-rich AGB star W Hya using data obtained by the HIFI, PACS, and SPIRE instruments on board the Herschel Space Observatory and ground-based telescopes.12 CO and 13 CO lines are used to constrain the intrinsic 12 C/ 13 C ratio from resolved HIFI lines. Methods. We combined a state-of-the-art molecular line emission code and a dust continuum radiative transfer code to model the CO lines and the thermal dust continuum. Results. The acceleration of the outflow up to about 5.5 km s −1 is quite slow and can be represented by a β-type velocity law with index β = 5. Beyond this point, acceleration up the terminal velocity of 7 km s −1 is faster. Using the J = 10-9, 9-8, and 6-5 transitions, we find an intrinsic 12 C/ 13 C ratio of 18 ± 10 for W Hya, where the error bar is mostly due to uncertainties in the 12 CO abundance and the stellar flux around 4.6 μm. To match the low-excitation CO lines, these molecules need to be photo-dissociated at ∼500 stellar radii. The radial dust emission intensity profile of our stellar wind model matches PACS images at 70 μm out to 20 (or 800 stellar radii). For larger radii the observed emission is substantially stronger than our model predicts, indicating that at these locations there is extra material present. Conclusions. The initial slow acceleration of the wind may imply inefficient dust formation or dust driving in the lower part of the envelope. The final injection of momentum in the wind might be the result of an increase in the opacity thanks to the late condensation of dust species. The derived intrinsic isotopologue ratio for W Hya is consistent with values set by the first dredge-up and suggestive of an initial mass of 2 M or more. However, the uncertainty in the isotopologic ratio is large, which makes it difficult to set reliable limits on W Hya's main-sequence mass.