High-resolution spectra of Pluto in the 1.66 μm region, recorded with the VLT/CRIRES instrument in 2008 (2 spectra) and 2012 (5 spectra), are analyzed to constrain the spatial and vertical distribution of methane in Pluto's atmosphere and to search for mid-term (4 year) variability. A sensitivity study to model assumptions (temperature structure, surface pressure, Pluto's radius) is performed. Results indicate that (i) no variation of the CH4 atmospheric content (column-density or mixing ratio) with Pluto rotational phase is present in excess of 20%, (ii) CH4 column densities show at most marginal variations between 2008 and 2012, with a best guess estimate of a ∼20% decrease over this time frame. As stellar occultations indicate that Pluto's surface pressure has continued to increase over this period, this implies a concomitant decrease of the methane mixing ratio (iii) the data do not show evidence for an altitude-varying methane distribution; in particular, they imply a roughly uniform mixing ratio in at least the first 22-27 km of the atmosphere, and high concentrations of low-temperature methane near the surface can be ruled out. Our results are also best consistent with a relatively large (>1180 km) Pluto radius. Comparison with predictions from a recently developed global climate model indicates that these features are best explained if the source of methane occurs in regional-scale CH4 ice deposits, including both low latitudes and high Northern latitudes, evidence for which is present from the rotational and secular evolution of the near-IR features due to CH4 ice. Our "best guess" predictions for the New Horizons encounter in 2015 are: a 1184 km radius, a 17 μbar surface pressure, and a 0.44% CH4 mixing ratio with negligible longitudinal variations.