A parameterized model of the mass distribution within the Milky Way is fitted to the available observational constraints. Constraints at R < R0 and at R ≃ R0 are well satisfied, while those at 1 < R/R0 <∼ 2 cannot be adequately fitted with the consequence that the circular speed vc(R) at R > R0 is extremely uncertain. The most important single parameter is the ratio of the scale length Rd,∗ of the stellar disk to R0. The disk and bulge dominate vc(R) at R <∼ R0 only for Rd,∗/R0 <∼ 0.25. Since the only knowledge we have of the halo derives from studies like the present one, we allow it to contribute to the density at all radii. When allowed this freedom, however, the halo causes changes in assumptions relating to R ≪ R0 to affect profoundly the structure of the best-fitting model at R ≫ R0, and vice versa. For example, changing the disk slightly from an exponential surface-density profile significantly changes the form of vc(R) at R ≫ R0, where the disk makes a negligible contribution to vc. Moreover, minor changes in the constraints can cause the halo to develop a deep hole at its centre that is not physically plausible. These problems call into question the proposition that flat rotation curves arise because galaxies have physically distinct halos rather than outwards-increasing mass-to-light ratios. The mass distribution of the Galaxy and the relative importance of its various components will remain very uncertain until more observational data can be used to constrain mass models. Data that constrain the Galactic force field at z >∼ R and at R > R0 are especially important.